Abstract
Optimizing short-term decisions over a rolling horizon and/or using deterministic penalties to capture system stochasticity can lead to myopic policies that fail to consider unplanned events and their long-term adverse effects. We present a methodology that integrates an off-line optimization model with a simulation procedure to determine the profitability of different operating strategies; specifically, the latter is used to generate additional constraints for the former when failures occur according to system component operating lifetimes that (i) are subject to exogenous uncertainty, and (ii) may degrade more quickly under specific operating conditions. We use the feedback provided by the simulation model in a parametric analysis to obtain penalties that can be used in short-term operations scheduling to maximize the long-term revenues obtained by the optimization model. We apply this research to a concentrating solar power plant; our results show that the methodology can be used to choose an operating policy that balances maximizing profit while accounting for maintenance costs. Integrating the optimization model with a simulation procedure reveals that aggressive prices for cycling yield about 55% fewer startups and 30% fewer failures compared to using a more typical start-up operating strategy, and can save hundreds of thousands to millions of dollars in repair costs over the lifetime of the plant.
Similar content being viewed by others
Data availability
We have supplied all available data and associated links in the paper itself. Other data is provided by our industry partners and is confidential.
References
Aguilar O, Kim JK, Perry S, Smith R (2008) Availability and reliability considerations in the design and optimisation of flexible utility systems. Chem Eng Sci 63(14):3569–3584
Alagoz O, Hsu H, Schaefer AJ, Roberts MS (2010) Markov decision processes: a tool for sequential decision making under uncertainty. Med Decis Mak 30(4):474–483
Altés Buch Q (2014) Dynamic modeling of a steam rankine cycle for concentrated solar power applications. PhD thesis, Polytechnic University of Catalonia, Barcelona, Spain
Aminov R, Kozhevnikov A (2017) Optimization of the operating conditions of gas-turbine power stations considering the effect of equipment deterioration. Therm Eng 64(10):715–722
Andradóttir S (2006) An overview of simulation optimization via random search. Handb Oper Res Manag Sci 13:617–631
April J, Glover F, Kelly JP, Laguna M (2003) Practical introduction to simulation optimization. In: Proceedings of the 35th conference on Winter simulation: driving innovation, Winter Simulation Conference, pp 71–78
Basciftci B, Ahmed S, Gebraeel NZ, Yildirim M (2018) Stochastic optimization of maintenance and operations schedules under unexpected failures. IEEE Trans Power Syst 33(6):6755–6765
Basciftci B, Ahmed S, Gebraeel N (2019) Data-driven maintenance and operations scheduling in power systems under decision-dependent uncertainty. IISE Trans pp 1–14
Benato A, Stoppato A, Bracco S (2014) Combined cycle power plants: a comparison between two different dynamic models to evaluate transient behaviour and residual life. Energy Convers Manag 87:1269–1280
Blair N, DiOrio N, Freeman J, Gilman P, Janzou S, Neises T, Wagner M (2017) System Advisor Model, SAM 2017.9.5: general description. Tech. Rep., National Renewable Energy Laboratory
Borgonovo E, Marseguerra M, Zio E (2000) A Monte Carlo methodological approach to plant availability modeling with maintenance, aging and obsolescence. Reliab Eng Syst Saf 67(1):61–73
Chang D, Jiang Z, Yan W, He J (2011) Developing a dynamic rolling-horizon decision strategy for yard crane scheduling. Adv Eng Inform 25(3):485–494
Chen CH, Fu MC, Shi L (2008) Simulation and optimization. In: State-of-the-art decision-making tools in the information-intensive age, pp 247–260
Ciocan DF, Farias V (2012) Model predictive control for dynamic resource allocation. Math Oper Res 37(3):501–525
Correa-Posada CM, Sánchez-Martín P (2013) Stochastic contingency analysis for the unit commitment with natural gas constraints. In: 2013 IEEE Grenoble Conference, IEEE, pp 1–6
Eide S, Wierman T, Gentillon C, Rasmuson D, Atwood C (2007) Industry-average performance for components and initiating events at US commercial nuclear power plants. Idaho National Laboratory, US Nuclear Regulatory Commission Office of Nuclear Regulatory Research Washington, DC p 20555
Fishman GS (1996) Monte Carlo: concepts, algorithms, and applications
Frangopoulos CA, Dimopoulos GG (2004) Effect of reliability considerations on the optimal synthesis, design and operation of a cogeneration system. Energy 29:309–329
Fu MC, Glover FW, April J (2005) Simulation optimization: a review, new developments, and applications. Proc Winter Simul Conf 2005:83–95
Gilchrist W (1993) Modelling failure modes and effects analysis. Int J Qual Reliab Manag 10(5)
Guédez R, Spelling J, Laumert B (2015) Reducing the number of turbine starts in concentrating solar power plants through the integration of thermal energy storage. J Solar Energy Eng 137(1):011003
Haghifam MR, Manbachi M (2011) Reliability and availability modelling of combined heat and power (CHP) systems. Int J Elec Power Energy Syst 3(33):385–393
Hale P, Arno R (2001) Survey of reliability and availability information for power distribution, power generation, and HVAC components for commercial, industrial, and utility installations. IEEE Trans Ind Appl 37(1):191–196
Hazan E et al (2016) Introduction to online convex optimization. Found Trends Optim 2(3–4):157–325
Herrmann JW (2006) Handbook of production scheduling, vol 89. Springer, Heidelberg
Higle JL, Sen S (1991) Stochastic decomposition: an algorithm for two-stage linear programs with recourse. Math Oper Res 16(3):650–669
IEEE (2007) IEEE recommended practice for the design of reliable industrial and commercial power systems
Islam MT, Huda N, Abdullah A, Saidur R (2018) A comprehensive review of state-of-the-art concentrating solar power (CSP) technologies: current status and research trends. Renew Sustain Energy Rev 91:987–1018
Jaillet P, Wagner MR (2012) Online optimization. Springer, Heidelberg
Jung JY, Blau G, Pekny JF, Reklaitis GV, Eversdyk D (2004) A simulation based optimization approach to supply chain management under demand uncertainty. Comput Chem Eng 28(10):2087–2106
Keatley P (2014) Cost modelling of coal power plant start-up in cyclical operation. In: Coal Power Plant Materials and Life Assessment, pp 358–388
Keatley P, Shibli A, Hewitt N (2013) Estimating power plant start costs in cyclic operation. Appl Energy 111:550–557
Kim SH, Nelson BL (2006) Selecting the best system. Handb Oper Res Manag Sci 13:501–534
Kleijnen JP (2008) Response surface methodology for constrained simulation optimization: an overview. Simul Model Pract Theory 16(1):50–64
Kumar N, Besuner P, Lefton S, Agan D, Hilleman D (2012) Power plant cycling costs. Tech Rep Natl Renew Energy Lab 303:275–3000
Kumar R, Sharma A, Tewari P (2012) Markov approach to evaluate the availability simulation model for power generation system in a thermal power plant. Int J Indus Eng Comput 3(5):743–750
Kushner H, Yin GG (2003) Stochastic approximation and recursive algorithms and applications, vol 35. Springer, Heidelberg
Law AM, Kelton WD, Kelton WD (1991) Simulation modeling and analysis, vol 2. McGraw-Hill, New York
Le K, Day J (1982) Rolling horizon method: a new optimization technique for generation expansion studies. IEEE Trans Power Appar Syst 9:3112–3116
Liu Y, Li M, Lian H, Tang X, Liu C, Jiang C (2018) Optimal dispatch of virtual power plant using interval and deterministic combined optimization. Int J Elec Power Energy Syst 102:235–244
Majidi-Qadikolai M, Baldick R (2016) Stochastic transmission capacity expansion planning with special scenario selection for integrating \( n-1\) contingency analysis. IEEE Trans Power Syst 31(6):4901–4912
Mak WK, Morton DP, Wood RK (1999) Monte Carlo bounding techniques for determining solution quality in stochastic programs. Oper Res Lett 24(1–2):47–56
Marseguerra M, Zio E (2000) Optimizing maintenance and repair policies via a combination of genetic algorithms and Monte Carlo simulation. Reliab Eng Syst Saf 68(1):69–83
Menéndez RP, Martínez JA, Prieto MJ, Barcia LA, Sánchez JMM (2014) A novel modeling of molten-salt heat storage systems in thermal solar power plants. Energies 7:6721–6740
Mirandola A, Stoppato A, Lo Casto E (2010) Evaluation of the effects of the operation strategy of a steam power plant on the residual life of its devices. Energy 35:1024–1032
Olafsson S, Li X, Wu S (2008) Operations research and data mining. Eur J Oper Res 187(3):1429–1448
Panneton F, L’ecuyer P, Matsumoto M (2006) Improved long-period generators based on linear recurrences modulo 2. ACM Trans Math Softw 32(1):1–16
Powell KM, Edgar TF (2012) Modeling and control of a solar thermal power plant with thermal energy storage. Chem Eng Sci 71:138–145
Powell WB (2007) Approximate dynamic programming: solving the curses of dimensionality, vol 703. Wiley, Oxford
Ramezanian R, Saidi-Mehrabad M, Fattahi P (2013) MIP formulation and heuristics for multi-stage capacitated lot-sizing and scheduling problem with availability constraints. J Manuf Syst 32(2):392–401
Sabouhi H, Abbaspour A, Fotuhi-Firuzabad M, Dehghanian P (2016) Reliability modeling and availability analysis of combined cycle power plants. Int J Elec Power Energy Syst 79:108–119
Santos MI, Uturbey W (2018) A practical model for energy dispatch in cogeneration plants. Energy 151:144–159
Shah D, Wischik D (2012) Switched networks with maximum weight policies: fluid approximation and multiplicative state space collapse. Ann Appl Probab 22(1):70–127
Silvente J, Kopanos GM, Pistikopoulos EN, Espuña A (2015) A rolling horizon optimization framework for the simultaneous energy supply and demand planning in microgrids. Appl Energy 155:485–501
Stoppato A, Mirandola A, Meneghetti G, Lo Casto E (2012) On the operation strategy of steam power plants working at variable load: technical and economic issues. Energy 37:228–236
Tică A, Guéguen H, Dumur D, Faille D, Davelaar F (2012) Design of a combined cycle power plant model for optimization. Appl Energy 98:256–265
Wagner MJ, Newman AM, Hamilton WT, Braun RJ (2017) Optimized dispatch in a first-principles concentrating solar power production model. Appl Energy 203:959–971
Wagner MJ, Hamilton WT, Newman A, Dent J, Diep C, Braun R (2018) Optimizing dispatch for a concentrated solar power tower. Solar Energy 174:1198–1211
Wu L, Shahidehpour M, Li T (2007) Stochastic security-constrained unit commitment. IEEE Trans Power Syst 22(2):800–811
Yildirim M, Sun XA, Gebraeel NZ (2016a) Sensor-driven condition-based generator maintenance scheduling-Part I: maintenance problem. IEEE Trans Power Syst 31(6):4253–4262
Yildirim M, Sun XA, Gebraeel NZ (2016b) Sensor-driven condition-based generator maintenance scheduling-Part II: incorporating operations. IEEE Trans Power Syst 31(6):4263–4271
Zare V (2016) Exergoeconomic analysis with reliability and availability considerations of a nuclear energy-based combined cycle power plant. Energy 96:187–196
Acknowledgements
This work is funded by the United States Department of Energy under award number DE-EE00030338. The authors appreciate the support of Jolyon Dent, formerly at SolarReserve\(^{\circledR }\), for help in modeling the CSP power cycle components. The authors thank Ernie Kee, of the University of Illinois at Urbana-Champaign, for his help in obtaining data sources for component failure rates, and Professor Dave Morton of Northwestern University for his helpful comments on prior drafts.
This paper was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest. That is, the authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Dispatch Formulation
Appendix: Dispatch Formulation
This appendix presents, verbatim, the dispatch model formulation (\(\mathcal {O}^\mathcal {R}\)), as summarized in the appendix of Wagner et al. (2018). Units, where appropriate, are provided in brackets next to the corresponding parameter or variable. (Initialization parameters used to set variable values at \(t=0\) subscribe to variable notation and are not included here.)
Indices and sets | |
\(t \in \mathcal {T}\) | Set of all time steps in the time horizon, \(T = |\, \mathcal {T}\ |\) |
Parameters | |
\(C^{chs}\) | Penalty for cycle start-up (from hot idle) [$] |
\(C^{csu}\) | Penalty for cycle start-up (from 0) [$] |
\(C^{\delta W}\) | Penalty for a positive change in electricity production [$/kWe] |
\(C^{rsu}\) | Penalty for receiver start-up (from 0) [$] |
\(E^c\) | Required energy consumed to start cycle [kWht] |
\(E^{hs}\) | Heliostat field startup parasitic loss [kWhe] |
\(E^r\) | Required energy consumed to start receiver [kWht] |
\(E^{rsb}\) | Tower piping heat trace parasitic loss [kWhe] |
\(E^u\) | Energy storage capacity [kWht] |
\(L^c\) | Cycle pumping power per unit energy consumed [kWe/kWt] |
\(L^r\) | Receiver pumping power per unit power produced [kWe/kWt] |
\(\mathbb {M}\) | A sufficiently large number [-] |
\(P_t\) | Electricity sales price in time t [$/kWhe] |
\(\bar{P}\) | Mean sales price \(\sum _{t \in \mathcal {T}} P_t / T\); [$/kWhe] |
\(Q^b\) | Standby thermal power consumption per period [kWt] |
\(Q^c\) | Allowable power per period for cycle start-up [kWt] |
\(Q^{in}_t\) | Energy generated by the solar field in time t [kWt] |
\(Q^l\) | Minimum operational thermal power input to cycle [kWt] |
\(Q^{rl}\) | Minimum operational thermal power delivered by receiver [kWt] |
\(Q^{ru}\) | Allowable power per period for receiver start-up [kWt] |
\(Q^u\) | Cycle thermal power capacity [kWt] |
\(W^b\) | Power cycle standby operation parasitic load [kWe] |
\(W^h\) | Heliostat field tracking parasitic loss [kWe] |
\(W^l\) | Minimum electric power output from cycle [kWe] |
\(\dot{W}^{min}_t\) | Minimum net power production in time t [kWe] |
\(\dot{W}^{net}_t\) | Net power production upper limit in time t [kWe] |
\(W^u\) | Cycle electric power rated capacity [kWe] |
\(\alpha \) | Unit corrector for binary variables in objective [$] |
\(\gamma _t\) | Exponential time weighting factor [-]; \(\varGamma ^{(t)}\), where \(\varGamma \approx 0.99\) |
\(\varDelta \) | Time step duration [hr] |
\(\varDelta ^l\) | Minimum duration of receiver start-up in period [hr] |
\(\varDelta ^{rs}_t\) | Estimated fraction of time step t used for receiver start-up [-] |
\(\eta ^{amb}_t\) | Cycle efficiency adjustment factor in time t [-] |
\(\eta ^c_t\) | Normalized condenser parasitic loss in time t [-] |
\(\eta ^{des}\) | Cycle nominal efficiency [-] |
\(\eta ^{p}\) | Slope of linear approximation of power cycle performance curve [-] |
Continuous variables | |
\(x^r_t\) | Thermal power delivered by the receiver at time t [kWt] |
\(x^{rsu}_t\) | Receiver start-up power consumption at time t [kWt] |
\(u^{rsu}_t\) | Receiver start-up energy inventory at time t [kWht] |
\(x_t\) | Cycle thermal power consumption at time t [kWt] |
\(u^{csu}_t\) | Cycle start-up energy inventory at time t [kWht] |
\(\dot{w}_t\) | Electrical power generation at time t [kWe] |
\(\dot{w}^{\delta }_t\) | Positive change in electricity production at time t [kWe] |
\(s_t\) | Thermal energy storage reserve quantity at time t (auxiliary variable) [kWht] |
Binary variables | |
\(y^r_t\) | 1 if receiver is generating “usable” thermal power at time t; 0 otherwise |
\(y^{rsu}_t\) | 1 if receiver is starting up at time t; 0 otherwise |
\(y^{rsup}_t\) | 1 if receiver incurs a penalty for start-up at time t; 0 otherwise |
\(y_t\) | 1 if cycle is generating electric power at time t; 0 otherwise |
\(y^{csu}_t\) | 1 if cycle is starting up at time t; 0 otherwise |
\(y^{csb}_t\) | 1 if cycle is in standby mode at time t; 0 otherwise |
\(y^{csup}_t\) | 1 if cycle is starting up at time t from off state; 0 otherwise |
\(y^{chsp}_t\) | 1 if cycle is starting up at time t from standby mode; 0 otherwise |
The objective function and constraints follow:
subject to
(Receiver start-up)
(Receiver supply and demand)
(Cycle start-up)
(Power supply and demand)
(Logic governing cycle modes)
(Energy balance)
(Variable bounds)
We maximize the product of electricity price and power generation less parasitic losses and cost penalties; this represents revenue associated with electricity sales minus total costs, and each of these differences is summed over the time horizon. The final term in the objective function incentivizes energy dumping from the solar field – if necessary – to occur later in the time horizon to improve agreement between the expected net electricity production and the actual modeled value, as pumping parasitics associated with solar field operation are significant.
Constraint (A.2a) accounts for start-up energy “inventory”; we use an inequality to permit a zero value in time periods following completion of a start-up. Constraint (A.2b) ensures that inventory can only be positive when the receiver starts. Constraint (A.2c) allows the receiver to produce power under only one of the following two circumstances: (i) after the completion of a start-up or (ii) if, in the previous time step, the receiver had been operating. Constraint (A.2d) ensures that a receiver cannot be starting up and operating in power-producing mode simultaneously. Constraint (A.2e) limits the power used for receiver start-up based on the ramp rate. Constraint (A.2f) precludes receiver start-up in time periods with insufficient solar resource.
There is a limit on the total power produced by the receiver, which is adjusted for start-up energy consumption by Constraint (A.3a). Constraint (A.3b) ensures that the receiver must be in power-producing mode to generate thermal power. There is also a minimum limit on receiver energy generation by Constraint (A.3c) to adhere to molten-salt pump operating limits and heat transfer requirements. Constraint (A.3d) precludes power from being produced in the absence of available energy. Constraint (A.3e) governs logic associated with incurring a receiver start-up.
Inventory associated with start-up energy is accounted for by Constraint (A.4a); when the cycle is starting up, inventory can assume a positive value by Constraint (A.4b). Typical operation of the cycle can occur only after completion of start-up, or if the cycle had been operating or had been in stand-by mode in the previous time step, by Constraint (A.4c). Use of thermal power during cycle start-ups is curtailed in Constraint (A.4d), and thermal power use by the power cycle is restricted to a maximum value by Constraint (A.4e). Constraint (A.4f) ensures a lower limit on power production when the cycle is operating.
Electrical power production is given by a constraint that relates cycle performance to thermodynamic efficiency, allowing the model to capture the relationship between efficiency and thermal input while maintaining a linear relationship between the two. Constraint (A.5a) exploits this relationship; the power cycle does not always operate at maximum efficiency owing to other cost considerations accounted for by the model. To resolve the nonlinearities present in the relationship between thermodynamic efficiency and thermal input, we approximate electrical output with a linear function of cycle thermal power consumption, as given in Constraint (A.5a) where
Constraint (A.5b) determines a positive change in electrical power production in any given time step, which is discouraged in the objective function. Electrical power production is subject to a lower bound in Constraint (A.5c) which, if it lies below the minimum cycle power value, forces the cycle to be turned off; see Constraint (A.5d). We refer the reader to §2.4.4 of Wagner et al. (2017) for more detail.
Start-up mode cannot occur if the cycle is operating by Constraint (A.6a). Constraint (A.6b) enforces the same type of restriction for standby mode. The modes of standby and start-up cannot simultaneously occur (Constraint (A.6c)), nor can the modes of standby and power production (Constraint (A.6d)). Penalties of start-up from an off or standby state are incurred based on logic implemented in Constraints (A.6e) and (A.6f), respectively.
The charge state of thermal storage (\(s_t\)) is equal to the difference between the corresponding positive and negative amount of power. Related constraints enforce energy balance with respect to thermal energy storage, where the parameter \(\varDelta \) converts from power to energy; see Constraint (A.7a). We account for the hourly time resolution of this model and the possible sub-hourly amount of time required to start the receiver while the power cycle is operating with Constraint (A.7b). If applicable, Equation (A.10) represents the fraction of each time step used for receiver start-up.
Thermal energy storage state of charge is calculated in every time step by Constraints (A.7a)-(A.7b).
Constraint (A.8a) imposes nonnegativity on receiver start-up power consumption and receiver start-up energy inventory. Constraint (A.3c) guarantees nonnegativity for \(x^r_t\). Nonnegativity for cycle start-up energy inventory, electrical power generation, and positive change in electricity production is given by Constraint (A.8b). Nonnegativity for \(x_t\) is ensured via Constraint (A.4f). Constraint (A.8c) bounds thermal energy storage in each time step. Constraints (A.8d) and (A.8e) enforce binary requirements.
Rights and permissions
About this article
Cite this article
Wales, J.G., Zolan, A.J., Hamilton, W.T. et al. Combining simulation and optimization to derive operating policies for a concentrating solar power plant. OR Spectrum 45, 119–150 (2023). https://doi.org/10.1007/s00291-022-00688-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00291-022-00688-7