Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid CONCLUDING REPORT Systems Engineering Institute Xi’an Jiaotong University Dec 9 2011 2 Modification History Version Date Name Contents 0.1 Nov 26 2011 Hao Zhang Create the document 0.2 Nov 30 2011 Hao Zhang Add the electricity pricing formation mechanism 0.4 Dec 4 2011 Hao Zhang Add the test result of typical case 0.5 Dec 6 2011 Kun Liu 0.6 Dec 8 2011 Hao Zhang Add the analysis of dynamic pricing problem 0.7 Dec 10 2011 Hao Zhang Add the solution of dynamic pricing problem 0.7 Dec 12 2011 Siyun Chen Modify the operational mode of mechanism 0.8 Dec 13 2011 Hao Zhang Add the general dynamic pricing mechanism 0.8 Dec 15 2011 Hao Zhang Add the background 0.9 Dec 18 2011 Hao Zhang Modify the symbol description 1.0 Dec 18 2011 Hao Zhang Integrate and modify all document parts 1.1 Dec 19 2011 Hao Zhang Add the expended test and analysis 1.2 Dec 21 2011 Hao Zhang Add the abstract and conclusion 2.0 Dec 23 2011 Hao Zhang Modify all document parts 2.0 Dec 23 2011 Zhaojie Wang Modify all document parts 2.1 Dec 26 2011 Hao Zhang Modify the document based on amendments 2.2 Jan 6 2012 Hao Zhang Modify the chapter of numerical test 3.0 Jan 9 2012 Zhaojie Wang 3.1 Jan 11 2012 Hao Zhang Add the response analysis and symbol description Modify the whole document Add a test result to typical case and conclusion 4 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Abstract The dynamic pricing problem in microgrid for energy intensive enterprise (EIE) is studied in this project. We focus on a microgrid which has distributed energy resources, energy storage devices and utility electricity. It has particular power consumption characteristic: surge-type load caused by batch production, complicated time-coupling and space-coupling constraints of end users. The scope of work is as following: 1) Analysis of elastic power demand and rigid power demand in EIE microgrid, time coupling relationship analysis among production units; 2) Analysis of how real-time power cost is affected by change of electrical load of production units; 3) Dynamic pricing models considering enterprise power purchase cost, enterprise generating cost, effect of production units on real-time power cost and other factor; 4) By typical case study, put forward some reference ideas about management of electrical utilization and generation for EIE microgrid. Base on the research, practical models for describing influence of price maker and price taker to real-time power cost are built. We also establish three dynamic pricing mechanism considering different types of production units’ electrical utilization, unit output, energy storages and time coupling relationship among production units. And a systematic guiding ideology about management of electrical utilization and generation for EIE microgrid is given. I Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises II Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid CONTENTS 1 2 3 4 5 6 7 Abstract .................................................................................................................. I Background ............................................................................................................ 1 1.1 Background ................................................................................................. 1 1.2 Dynamic Pricing Problem in Industrial Microgrid...................................... 3 1.3 Target ........................................................................................................... 3 1.4 Method Introduction .................................................................................... 4 Symbol Description ............................................................................................... 4 The Analysis of Dynamic Pricing Problem ........................................................... 6 3.1 The Participants of Dynamic Pricing Mechanism ....................................... 6 3.2 Some Issues Involved .................................................................................. 6 The Solution ........................................................................................................... 7 4.1 The Operational Mode of Dynamic Pricing Mechanism ............................ 7 4.2 The Analysis of Dynamic Pricing Mechanism ............................................ 8 4.3 Principles to follow in solution ................................................................... 8 4.4 Problem Decomposition and Analysis ........................................................ 8 Response Analysis of Power Consumption and Generation .................................. 9 5.1 Classification of Electricity Consumption Equipments and Storage Devices ..................................................................................................................... 9 5.1.1 Classification of Batch Production Equipment ................................. 9 5.1.2 Classification of Continuous Production Equipment...................... 10 5.1.3 Classification of Storage Equipments ............................................. 10 5.2 Power Consumption Optimization Model ................................................. 11 5.2.1 Batch Load Curve Analysis ............................................................ 11 5.2.2 The Load Management Model ........................................................ 12 5.3 Power Generation Optimization Model .................................................... 13 5.4 Use Case .................................................................................................... 14 The Electricity Pricing Formation Mechanism .................................................... 17 6.1 The Brief Introduction of Real-time Cost Analysis................................... 17 6.2 Optimal Power Generation Strategy.......................................................... 17 6.2.1 The Cost Analysis of Generation Strategies ................................... 17 6.2.2 The Formulation of Optimal Power Generation Strategy ............... 18 6.3 Pricing Formation Mechanism for Energy Intensive Enterprise Microgrid .. ................................................................................................................... 20 6.3.1 Real-time Cost Based Price Formation Mechanism ....................... 20 6.3.2 Gradient Information Based Price Formation Mechanism ............. 22 6.3.3 Heuristic Method Based Price Formation Mechanism ................... 24 Test Result of Typical Cases and the Analysis ..................................................... 27 7.1 Cases Introduction ..................................................................................... 27 7.2 Pricing Formation Mechanism Based on Real-time Cost ......................... 29 7.3 Pricing Formation Mechanism Based on Gradient Information ............... 31 III Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 7.4 7.5 Pricing Formation Mechanism Based on Heuristic Information ............... 33 Expended Test and the Analysis ................................................................ 36 7.5.1 Change the Batch Number: ............................................................. 36 7.5.2 Analysis of Cost-saving .................................................................. 40 7.5.3 Change the Range of Units Output ................................................. 42 8 Conclusion ........................................................................................................... 53 9 The Description of General Dynamic Pricing Mechanism in Microgrid ............ 55 Reference .................................................................................................................... 58 Equation Chapter 1 Section 1 IV Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 1 Background 1.1 Background Generally, dynamic pricing refers to a type offer or contract by a provider of a service or supplier of a commodity, in which the price depends on the time when the service is provided or the commodity is delivered. In electricity market, dynamic pricing, as a demand response, provides customers with time-varying prices that reflect wholesale market costs. Dynamic pricing rate fluctuations follow the real time cost of electricity. The objective of these programs is to flatten the demand curve by offering a high price during peak periods and lower prices during off-peak periods. Most common rates are Time of Use (TOU), Critical Peak Pricing (CPP), and Real Time Pricing (RTP).[1] The power supply pattern of industries without self generation power plant is shown in Figure 1-1. Use less energy during peak hours, or move the time of energy use to off-peak times under a specified electricity tariff like TOU implemented by the utilities do help industries reduce their electricity costs. The Utility Demand Side Energy management section User 1 Figure 1-1 User M User 2 without self generation power plant Electricity distribution 1 0.9 0.8 Price/¥ /kwh 0.7 0.6 0.5 0.4 0.3 0.2 Price of buying electricity Price of selling electricity 0.1 0 0 20 40 Figure 1-2 60 80 Time/10min 100 120 140 the TOU tariff given by utility Most energy intensive enterprises (EIEs) have self generation power plant. Based 1 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises on satisfying production related objectives, reducing electricity costs becomes an important objective for production scheduling in EIEs. Such an EIE integrates power generation, power consumption and energy storage into a unified whole, and become an enterprise microgrid in Figure 1-3. Also, the structure is highly coupled with a multi-energy system which is formed by production processes and other energy medium. The Utility Microgird Generator 1 Generator Generator … N 2 Energy management section User 1 Figure 1-3 User 2 User M … the power supply pattern of microgrid For such an EIE microgrid, electricity cost is concerned with prices of buy/sell electricity from/to the utility, generating cost and load demand in production progress. The total cost dynamically changes the power demand and the power generation, and is inconsistent with tariff implemented by the utility shown in Figure 1-4. In order to achieve the purpose of saving energy costs, the EIE can establish a price mechanism to guide the production scheduling. Through dynamic pricing (DP) mechanism, energy management section uses an internal electricity pricing strategy to release price and to guide the behaviors of production units to reduce energy cost. 4 7.5 The Cost Curve x 10 7 6.5 Cost/¥ 6 5.5 Gas addition 50KNm3/h Gas addition 100KNm3/h 5 Gas addition 150KNm3/h 4.5 4 3.5 Fixed price TOU 0 Figure 1-4 2 10 20 30 40 Time/h 50 60 70 80 the electricity consumption cost curve under different gas addition Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Dynamic pricing has not applied practically in industrial microgrid. At present, the analysis of demand response behavior and the analysis of relationship between power consumption and production schedule are still vacancies. A rational price signal which can guide the demand response is expected. And a dynamic pricing mechanism in industrial microgrid is needed. 1.2 Dynamic Pricing Problem in Industrial Microgrid The dynamic pricing problem in microgrid for EIE is studied in this project. We focus on a microgrid which has distributed energy resources, energy storage devices and utility electricity. It has particular power consumption characteristic: surge-type load caused by batch production, complicated time-coupling and space-coupling constraints of end users. The focus of the problem is shown in Figure 1-5 Figure 1-5 focus of dynamic pricing problem in microgrid for EIE There are several requirements for a rational dynamic pricing in microgrid for EIE. The dynamic price should bring cost benefit for both enterprise and each production unit. For the energy management section, the break even is needed. And the dynamic price should have usability for manufacture management and frontline workers. Compare to the utility, the dynamic pricing problem in EIE microgrid has some unique features: a) Self generation power plant is connected to the microgrid. Surge-type loads caused by batch production are common seen in EIE. And the EIE microgird cannot island from the grid; b) Some EIE has by-product storage devices like gas holder. The by-products like blast-furnace gas is used for power generation. By-product scheduling is coupled with power generation scheduling and production scheduling; c) Production units have to be rational and obey the short term scheduling of manufacture management. And the load scheduling of production units should satisfy production related limitations. d) Power demand of production units are usually time coupling and space coupling. Each unit’s fine-tuning range of power demand is limited. 1.3 Target 3 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises The target of this project is forming a dynamic pricing mechanism in enterprise microgrid which including: a) Dynamic pricing mechanism; b) Real-time or lead-ahead price signals for demands response in microgrid; c) Possible energy cost reduction with the proposed pricing mechanism and rational demand response; 1.4 Method Introduction In this project, there are two method play important roles. The power consumption optimization model are built and the mixed integer linear programming (MILP) is used for finding the optimal solution. And the linear programming (LP) is used for the power generation optimization model. Iterative method, heuristic method and gradient method are also been used in project. LP is a mathematical method for determining a way to achieve the best outcome (such as lowest cost) in a given mathematical model for some list of requirements represented as linear relationships. Linear programming is a specific case of mathematical programming. More formally, linear programming is a technique for the optimization of a linear objective function, subject to linear equality and linear inequality constraints. [2] If only some of the unknown variables are required to be integers, then the problem is called a MILP problem.[3] Equation Chapter (Next) Section 1 2 Symbol Description Indices: m n index of batch production equipments, m 1, 2, index of power generators, n 1, 2, , N . k index of time periods, k 1, 2, j index of batch equipment number, j 1, 2, i index of iteration number, i 1, 2, length of each time period. ,M . ,K . , Jm . ,I Parameters: d kbase base load at time period k kbuy , ksell electricity price of buy/sell from/to the utility at time period k, buy ,k sell ,k 0 4 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid kgene electricity price of self generation power plant at time period k. sm, j , sm, j fine-tuning range of start time sm , j pn maximum allowable ramp rate of power output rate of power generator n. pn , pn maximum and minimum allowable power output rate of power MD generator n maximum demand V ,V maximum and minimum allowable fuel-storage level at time period k Jm total batch number of batch production equipment m in K time periods Tm batch time of batch production equipment m dcm d m ,k load characteristic value of batch production equipment m, where d m,,k 0,1, 2, , Tm . Variables: pn ,k i average power output of power generator n at time period k in i-th iteration d mi,k average load of batch production equipment m at time period k in i-th iteration ki internal electricity price for production units in i-th iteration vk fuel storage quantity at time period k vk available fuel amount for power generation flow into the storage device at time period k vk fuel consumption for power generation at time period k ve , k emission amount of gas holder at time period k sm , j , em , j start/stop time of j-th batch of batch production equipment m. Equation Chapter (Next) Section 1 5 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 3 The Analysis of Dynamic Pricing Problem 3.1 The Participants of Dynamic Pricing Mechanism In the dynamic pricing problem, there are five important participants, including production units, self generation power plant, the utility, energy management section. All those participants can be classified into 3 classes: power demander, power supplier and energy management section. The production units can be regarded as the power demander. They need to finish their daily task of production, and also can do some scheduling for production processes. The purpose of production scheduling is to minimize the production time and costs, by telling a production facility when to make, with which staff, and on which equipment. The production units aim to guarantee the production, maximize the efficiency of the operation and reduce costs, including energy consumption cost. Self generation power plant and the utility are the power suppliers. Self generation power plant is those power plants which operate independent of wheeling to the utility. The EIE microgrid needs to buy electricity from the utility under a certain tariff while self generation cannot meet its requirement. Thus, one of the responsibilities for self generation power plant is minimizing the total electricity cost of enterprise. The energy management section is the core participant of this mechanism. In position of the entire enterprise, the energy management section needs to use the price signal or scheduling method for minimizing energy cost, including generation cost, net electricity cost, penalty fees for gas emission, and energy storage fees, et al. The Utility Self Generation Power Plant Power message Tariff Price signal/ direct control Energy Management Section (Dynamic Pricing Mechanism) Price Load message signal Production unit Production unit …… Production unit Production units organize production considering electricity cost information Figure 3-1 the participants of dynamic pricing mechanism 3.2 Some Issues Involved 6 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid In a general context, the key points need to be focused on in dynamic pricing problem of industrial microgird are as follows: a) Production processes; the analysis of production processes helps us definite the time coupling and space coupling among processes and the load characteristics. b) Division of departments responsibilities; the division of department responsibility definite the role of various departments and the they need to offer. c) Production scheduling; when the dynamic pricing is used, the production scheduling helps to find out what kind of production plan can help the process unit minimize its cost. d) Generation scheduling; when the dynamic pricing is used, the generation scheduling helps to find out what kind of generation plan can help the whole enterprise minimize the total cost. e) Dynamic pricing method; the research on dynamic pricing method help us to make a reasonable internal tariff. And this tariff is suitable for a price signal and convenient for internal accounting.Equation Chapter (Next) Section 1 4 The Solution 4.1 The Operational Mode of Dynamic Pricing Mechanism In order to realize the dynamic pricing in industrial microgrid, a dynamic pricing mechanism is proposed. The operational mode of proposed dynamic pricing mechanism is shown in Figure 4-1. Energy Intensive Industrial Microgrid Production Management Department Initial feasible region for production scheduling Price Production Unit 1 Production Unit 2 Figure 4-1 …… Load Message Production Unit M The Utility Energy Management Section Price or Power Direct Control Message Price Load/Power Message Autonomous Power Plant the operational mode of dynamic pricing mechanism According to the initial feasible region of production scheduling provided by the production management section the energy, the energy management can formulate a dynamic price based on the price formation mechanism, though the analysis of the real-time electricity cost and the utility tariff, et al. 7 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Then the energy management section releases the dynamic electricity price signal to all users, so that users can perceive the current electricity costs and predict the influence of its electricity variation on their own real-time electricity cost. Based on the guidance of dynamic price, the mechanism encourages users to do power management spontaneously. And it lets the users to organize their production schedule according to the quantity, the rigidity of tasks and the electricity cost. And corresponding load messages are sent to energy management section. It helps the section give a price or direct control to the self generation power plant based on a certain generation strategy. The energy management section can get the power consumption of each user. A user m should pay its electricity fee according to the internal settlement cost, which is K computed by SCm k d m ,k . So the payment of all users is k 1 M K SC k d m ,k , which is equal to the total electricity cost of enterprise. m 1 k 1 In this way, every user is trying to reduce their cost. As a combination of individual users and generators, EIE can reduce the total electricity costs. 4.2 The Analysis of Dynamic Pricing Mechanism 4.3 Principles to follow in solution There are two principles to follow: cost-benefit for enterprise and usability for operators. The cost-benefit principle means to reduce electricity costs of energy intensive industrial micro grid by encouraging production units use less energy during peak hours, or move the time of energy use to off-peak times under a specified electricity tariff like TOU. And the real-time electricity cost analysis provides assessment criteria. The usability principle stands for the internal dynamic price mechanism should be easy to implement, and be simple for operators. That includes two aspects: TOU tariff is more suitable for giving industrial users a relatively stable price to rescheduling their production under a limited fine-tuning range; and by modeling customer behavior in price mechanism, the internal competition and optimization can be realized and set a final price for users as an internal price which is not a actual competition among production units. It releases the burden of production units in dynamic pricing mechanism. 4.4 Problem Decomposition and Analysis The dynamic pricing problem in industrial microgrid can be decomposed into 8 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid three parts: real-time analysis, electricity characteristic of equipments, and the price formation mechanism. In order to minimize the cost of the enterprise, the real-time cost should be analyzed to find out the composition and proportion of the cost. At the same time, the classification and analysis of equipments electricity characteristics are carried on. Based on the two steps above, the power consumption optimization model can be given which try to find out the optimal solution of minimizing the process units cost. By the power generation optimization model, we can get the optimal generator units output curve for minimal cost of enterprise. The optimal power consumption/ generation strategy is also needed which decide the generation strategy in different utility tariff and power consumption situation. At last, forming a useable dynamic price is the final goal, so the price formation mechanism should be designed. The relationship among all parts of the dynamic pricing problem is shown as follow:Equation Chapter (Next) Section 1 Dynamic pricing problem Real-time cost analysis Analysis and classification of equipment power consumption Power consumption optimization model Figure 4-2 Power generation optimization model Price formation mechanism Optimal power consumption/ generation strategy all parts of the dynamic pricing problem and their relationship 5 Response Analysis of Power Consumption and Generation 5.1 Classification of Electricity Consumption Equipments and Storage Devices 5.1.1 Classification of Batch Production Equipment Batch production equipments are related to the start time, the end time, the energy consumption per unit of time or the batch time. So batch production equipments can be divided into several types. But in these types, there are only 2 types existing. TABLE 5-1 classification of batch production equipment 9 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Batch time Energy consumption per unit of time adjustable nonadjustable adjustable Type 1 () () nonadjustable () Type 2 () Type 1: The start time of each batch, the batch time, and the energy consumption per unit of time are adjustable. For example, the electric furnace. Type 2: The start time of each batch is adjustable; the batch time and the energy consumption per unit of time are nonadjustable. For example, hot rolling and cold rolling. 5.1.2 Classification of Continuous Production Equipment Continuous production equipments are related to the start time, the end time, and energy consumption per unit of time. Then continuous production equipments can be divided into several types. But in these types, there are only 2 types existing. TABLE 5-2 classification of continuous production equipment Energy consumption per unit of time Begin time adjustable nonadjustable () Type 4 () Type 3 () () and end time adjustable nonadjustable Type 3: The start time and the end time are nonadjustable; the energy consumption per unit of time is adjustable. For example, the hot rolling and cold rolling. For example, the generator units. Type 4: The start time and the end time are adjustable; the energy consumption per unit of time is nonadjustable. For example, the blast furnace blowing. 5.1.3 Classification of Storage Equipments Storage cost is related to storage time and inventory level. And storage time and inventory level are functions of inflow, outflow and existing inventory. Then storage equipments can be divided into 4 types. But there are only 2 types existing. TABLE 5-3 Classification of intermediate storage devices due to composition of storage cost Inventory level Storage time related 10 related unrelated Type 1 () () Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid unrelated Type 2 () () Type 1: Storage cost is related to storage time and inventory level. For example, soaking pit. Type 2: Storage cost is only related to inventory level. For example, gasholder. 5.2 Power Consumption Optimization Model Participant of dynamic price mechanism is achieved by the optimization of equipment electricity program and the optimization of self generation power plant. 5.2.1 Batch Load Curve Analysis The load curve is directly related to production process. Production equipments can be divided into two types: batch production type, continuous production type. Unlike the quasi-periodicity and slow-variations load characteristics of main grid, the power consumption curve of EIE has the characteristics of serious surge. And the surge characteristic is caused by start/stop operation of large batch type production equipments which consumes a lot of electric energy. Thus, the total load of EIE can be decomposed into two parts including base load and batch load of large batch type production equipments. M Dk d m,k d kbase (5-1) m 1 Operating parameters and sequences of the batch process are shown in. Batch time 1 2 em , j sm , j sm , j em , j sm , j T em , j Time fine-tuning range Power demand Figure 5-1 operating parameters and sequences of the batch process The energy consumption of batch production equipments can be simulated by variable load factor. The load factor is the ratio between real load and rated load. Then the energy consumption of batch production equipments can be described as bellow: 11 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Energy consumption k-2 k-1 k k+1 k+2 Time It is possible that some batch production equipments may have several load curves. Then these equipments can be divided into several virtual equipments. These several virtual equipments have these own load curves. 5.2.2 The Load Management Model Objective Function The purpose of the load management is that by adjusting the production plan minimize the electricity cost of production. The objective is minimizing the electricity cost of production, formulated as bellow: Tm , j K min Cm dcm d m, j com sm, j d m, j CRm, j kcom d kbase (5-2) k 1 j 1 d m , j 1 Jm Constraint Condition The start of the j-th batch should be after the stop of the (j-1)-th batch for the mth equipment. sm, j 1 Tm sm , j (5-3) The last batch of the mth equipment should be stop previous to the end of one day. sm , j K ( J m j 1) Tm (5-4) The start of the jth batch of the mth equipment should be t time periods after the start of the jth batch of the mth equipment. sm ', j ' Tm ' tm ' j ',mj sm, j (5-5) where, tm ' j ', mj reflect factors like logistics speed between upstream and downstream equipments or timing constraints among batch production tasks of virtual equipments. 12 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Start time sm , j of batch j of production equipment m should be under fine-tuning time range: sm, j sm, j sm, j (5-6) where, sm, j , sm, j can also be used to describe timing constraints among batch production tasks of virtual equipments. Model solution The mixed integer linear programming (MILP) is used for model solution. 5.3 Power Generation Optimization Model The purpose of the power generation model is minimizing the electricity cost at one load curve. Objective Function The real-time variable electricity cost at time period k is represented by Ck , and is composed of fuel cost for self generation Gk and net electricity cost of buy/sell electricity from/ to the utility Bk , formulate as bellow: Gk Bk (5-7) Gk f n pn ,k (5-8) kbuyQk , if Qk 0 Bk 0, if Qk 0 sellQ , if Q 0 k k k (5-9) min Ck N n 1 Qk Lk Pk (5-10) where, Qk represent the net amount of power flow buy from utility at time period k, Lk represents the average total load of enterprise during time period k , Pk represent the average total power output of power generator during time period k, f n pn ,k represent the fuel cost function of power generator n during time period k. Constraint Condition Minimum/maximum generation output constraints: 13 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises pn pn , k pn (5-11) pn ,k 1 pn ,k pn (5-12) Ramp rate of generation output: Initial and terminal generation output: pn,0 pn*,0 , pn, K pn*, K (5-13) Balance between fuel supply and fuel demand for generator: vk vk 1 vk vk ve,k (5-14) v vk 1 vk V , if vk vk 1 vk V 0 ve,k k 0, else (5-15) Capacity of fuel storage device: V vk V (5-16) v0 v0* , vK vK* (5-17) Initial and terminal fuel storage: Model solution The linear programming (LP) method is used for model solution. 5.4 Use Case To illustrate the model, a case study for an iron and steel plant is conducted. For iron and steel industries, the electricity costs about 30% of the total production costs. In the context of increasing electricity prices and the introduction of time varying electricity rates by utilities, iron and steel plant can fine-tune their production operations schedule and schedule self generator units to reduce their electricity costs. The iron and steel enterprise being studied include three main factors that influence the real-time electricity costs: self gas-steam combination circulation generating unit (CCPP) with COREX furnace gas, time-of-use (TOU) tariff, production electricity loads. The production-gas-power correlation structure is shown as Fig. 1, where, the white arrow indicate the production processes, red arrows indicate electricity flow, and green arrows indicate gas flow; LF furnaces for steel making and rolling mills for medium plant production belong to the most highly energy intensive batch type production equipments; And self CCPP generator using COREX furnace gas provides 14 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid electricity for production process together with grid, power exchange between grid and enterprise is inevitable, electricity costs saving become a problem need to be considered under TOU tariff. Diffusion Other Use Gas Holder Oxygen Making COREX Furnace CCPP Steelmaking (LF Furnace) Production Process Medium Plate Production Electricity Flow Grid Gas Flow Fig. 1 production-gas-power correlation structure Parameter Setting and Related Data for Simulation We consider power generation, power consumption and power-related storage device in our use case study. Equipments and there parameters are list in Tab. I. TABLE I. No. Equipments EQUIPMENTS LIST AND PARAMETERS Type Parameters 1 LF furnace Batch production Power rating: 3000 (KW) 2 blooming mill Batch production Power rating: 3000(KW) 3 Finishing mill Batch production Power rating:5000 (KW) 4 CCPP generator Continuous generation Maximum output: 16945(KW) Minimum output:8500(KW) 5 Gas holder Storage device Storage capacity: 500000(Nm3); Lower limit: 160000(Nm3); Gas price:0.5(¥/ Nm3) Penalty of diffusion: 1(¥/ Nm3) The average power demands over every 10-min interval are noted and their variation is shown figure 2 for the batch time of LF furnace, blooming mill and finishing mill. 15 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Load curve 18 Active power Base load 17 16 Load/10000KW 15 14 13 12 11 10 9 8 0 50 100 150 Time/10min Figure 5-2 Load cycles of batch load and base load The power consumption optimization model can get the result as follow: Load curve 18 After optimization Before optimization 17 16 Load/10000KW 15 14 13 12 11 10 9 8 0 50 100 150 Time/10min Figure 5-3 the load cycles before and after power consumption optimization The power generation optimization model can get the result as follow: After optimization generator output and load 18 Generators out After optimization load 17 16 Load/10000KW 15 14 13 12 11 10 9 8 0 50 100 150 Time/10min Figure 5-4 16 load cycles and power generation output curve after optimization Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Equation Chapter (Next) Section 1 6 The Electricity Pricing Formation Mechanism 6.1 The Brief Introduction of Real-time Cost Analysis As we know, reducing the cost of whole EIE is one of the most important purposes of our electricity pricing mechanism. When forming the dynamic electricity price, the real-time cost must be fully used. And that makes the mechanism work at the direction of decreasing power consumption cost for the whole EIE during iteration. So analyzing the real-time cost of all kinds of equipments and the whole enterprise is required. Generally, power supply and demand are integrated in EIE microgrid. There are generator units, electricity consumption equipments and energy storage devices in the enterprise at the same time. The real-time costs in EIEs are related to the purchase/sale price from the utility, the real-time output of self generation power units and the internal real-time electricity load. Therefore, the electricity costs of EIEs are time-varying. The variable part of real-time electricity cost in EIE is variable cost, which includes generation fuel cost and electric energy exchange cost. For iron and steel enterprises, further influence factors of these variable costs are: output of generators, mixed burning proportion of generators, electricity load, gas price, electricity purchasing price, etc. These should be considered into the mathematical model. We also investigate the influence of a factor’s variance on cost in EIEs. More details about real-time electricity cost problem analysis, including the mathematical model, are described in document:” Real-time electricity cost problem analysis in EIE microgrid”. 6.2 Optimal Power Generation Strategy According to the analysis above, power supply and demand are integrated in EIE microgrid. That means there is an optimal power generation strategy for the power generation. The strategy determines the power proportion of generation and buying from utility, which attempts to minimize the power consumption cost. So the pricing formation mechanism needs to contain this strategy. 6.2.1 The Cost Analysis of Generation Strategies Considering the relationship between the load and units output, in a period k, the generation has three strategies: 17 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Case A: pk d k k Case B: pk d k (6-1) Case C: pk d k k Case A means the output is larger than the load at the k-th period. Case B shows that the output match the equipments load well at the k-th period. And case C means the units output is lower than the load. The three cases are shown in Figure 7-1. Electrical Quantity(KWh) A dk k B dk C dk k Time Period(h) Figure 6-1 the three cases of relationship between output and load At a same time k, there are three kind of price: price for buying, price for selling and the price for generation. The price for generation is denoted as kbuy , ksell and kgene . The real-time cost function under three cases can be described as: CkA kgene d k kgene ksell k CkB kgene d k (6-2) CkC kgene d k kbuy kgene k When the microgrid is in case A, generator output more power than need. The real-time cost of the enterprise is the cost of generation, plus the profit of selling electricity. When in case B, the real-time cost is only the cost of generation. And in case C, the real-time is the cost of generation, plus the cost of buying electricity from utility. Then we can formulate the strategy after the cost analysis under different price level. 6.2.2 The Formulation of Optimal Power Generation Strategy Based on the cost analysis of different generation strategies, we can get the cost curve shown in Figure 6-2. 18 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Ck CkC kgene d k kbuy kgene k CkA kgene d k kgene ksell k CkB kgene d k k Figure 6-2 cost curve in different strategies Consider the relationship among the 3 kinds of price, kbuy , ksell and kgene , three possible cases may happened, shown in Figure 6-3. kgene ksell Case 1 kgene kbuy Case 2 Figure 6-3 kgene k Case 3 the relationship of three kinds of price Then we can judge the minimal cost function, by the two figures above: In case 1: kbuy kgene , ksell kgene , then min CkA , CkB , CkC CkA . In case 2: kbuy kgene ksell , then min CkA , CkB , CkC CkB In case 3: kgene kbuy , kgene ksell , then min CkA , CkB , CkC CkC In electricity market, time of use (TOU) tariff is a kind of most common tariff. There are three general levels of TOU: peak price, valley price and flat price. Base on the analysis above, in order to minimize the cost of entire enterprise, we can formulate the optimal power generation strategy. There are three situations, shown in Figure 6-4: Situation 1: koff peak kgene , kgene ksell , kpeak kgene The optimal strategy: power generation follows power load while peak time; less power generation and buy more electricity from the utility while valley time. Situation 2: koff peak kgene , kgene ksell , kpeak kgene The optimal strategy: power generation follows power load all the time. Situation 3: koff peak kgene , kgene ksell , kpeak kgene The optimal strategy: power generation at full capacity. So the generator needs to follow this strategy to get minimal total cost. And this should be manifested in pricing formation mechanism. 19 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Situation three Situation two kgene Situation one peak 0 sell Figure 6-4 the 3 kinds of situation among TOU tariff and generation price off-peak Price (¥) 6.3 Pricing Formation Mechanism for Energy Intensive Enterprise Microgrid In order to obtain appropriate internal price, we need to design a price formation mechanism. The formed internal price can be treated as a signal to guide the user adjusted their production plan for reducing the total cost of the enterprise. 6.3.1 Real-time Cost Based Price Formation Mechanism As we know, reducing the cost of whole EIE is one of the most important purposes of our electricity pricing mechanism. So the real-time cost can be used directly to form a price. Based on the real-time cost analysis mentioned before, a real-time cost based price formation mechanism has been designed. We also call it the basic mechanism. The basic mechanism has the iteration base on real-time electricity cost and centralized power optimization of end users. The steps of the price formation mechanism are as follows: Step 1: the energy management section offers a initial price as an internal price at will, (0) k k 1, 2, , K , set i=1; Step 2: under the i-th internal price, all production units optimize their electricity consumption, to minimize their total power consumption cost, and get the d m ,k , i M K d mi,k arg min C m i,k m 1 k 1 k i 1 (6-3) Step 3: the self generation power plant does the centralized power generation optimization, and get the optimal units output, K pm i,k arg min Cki d mi,k ; (6-4) k 1 Step 4: using the generation cost, the cost of buying electricity and the total load, compute the real-time energy consumption cost price k , and i 20 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid M ki Cki / d mi,k ; (6-5) m 1 Step 5: if the total cost and price in this iteration doesn’t satisfy the condition of convergence: K i K i 1 K i Ck Ck / Ck C k 1 k 1 k 1 , i i 1 max k k / i k (6-6) then i=i+1, go to step 2. Step 6: output and release the internal price k as a dynamic price signal in i microgrid. And the flow chart of this real-time cost based price formation mechanism is shown in Figure 6-5. Energy management section offers a initial price k(0) k 1, 2, , K set i=1 i i 1 Electricity consumption optimization of production units under internal price M K d m ,k arg min C m ,k k i i i 1 m 1 k 1 Power generation optimization of self generation power plant K pmi ,k arg min Cki d mi,k , No k 1 Compute the real time energy consumption cost price M ki Cki / d mi,k m 1 Convergence? Yes Output the final price Figure 6-5 the flow chart of real-time cost based price formation mechanism This mechanism has several advantages: a) It manifests the important point of decreasing the total energy consumption cost of the whole enterprise intuitively; b) It doesn’t have special requirements to meet for initial price. The initial price can be given at will. 21 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises And it also has several disadvantages: a) Without using the historical , the iteration process of internal price is concussive, and hard to reach convergence; b) The final internal price is too weak to be used as a price signal; c) The final internal price has rapid fluctuation during 24 hours, so that it’s hard for practical use. And all the features mentioned above will be analyzed and explained in case study shown in next chapter. 6.3.2 Gradient Information Based Price Formation Mechanism Because basic price formation mechanism has several obvious drawbacks, such as the internal price is concussive and it’s difficult to use as a dynamic price. So this RTP tariff mechanism 1 has been designed. RTP tariff mechanism 1 is a kind of gradient information based price formation mechanism. And it has the iteration base on predefined price periods and centralized power optimization of end users. During iteration, the mechanism gives an increment to the price of the last iteration based on gradient information. The iteration direction needs to be the same as the gradient direction of the daily total cost, which can ensure reducing the cost of whole EIE. The gradient of the daily total cost is unknown. So considering the real-time cost analysis mentioned above, here we use the gradient of the k-th period cost instead. The increment function is chosen as follow: f Qki kbuy , if Qki 0 buy sell k k , if Qki 0 2 ksell , if Qki 0 (6-7) Where Qk means the in reception electricity in k-th period of i-th iteration. i The steps of the price formation mechanism are as follows: Step 1: the energy management section offers a initial price as an internal price at will, k 1, 2, 0 k , K , set i=1; Step 2: under the i-th internal price, all production units optimize their electricity consumption, to minimize their total power consumption cost, and get the d m ,k , i M K d mi,k arg min C m i,k m 1 k 1 k i 1 ; (6-8) Step 3: the self generation power plant does the centralized power generation 22 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid K optimization, and get the optimal units output, pmi,k arg min Cki d mi,k , k 1 Step 4: using the internal price generated by the last iteration, and the gradient function, compute the internal price k for next iteration, and the price i formulation is ki 1 k ki 1 k i M N m 1 n 1 f Qki , Qki d mi,k pmi,k ; f Qk i (6-9) kbuy , if Qki 0 buy sell k i k , if Qk 0 2 sell k , if Qki 0 Step 5: realizing the normalization through k k , to make sure that the i i i total internal settlement cost is consistent before and after the Step 4. K i C k 1 i k M a 1 f Qki ki 1 d mi,k i k 1 m 1 K (6-10) Step 6: if the total cost and price in this iteration doesn’t satisfy the condition of convergence: K i K i 1 K i Ck Ck / Ck C k 1 k 1 k 1 , (6-11) max ki ki 1 / ki then i=i+1, go to step 2. Step 7: output and release the internal price k as a dynamic price signal in i microgrid. And the flow chart of this gradient information based price formation mechanism is shown in Figure 6-6. 23 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Energy management section offers initial price and set i=1 k 1, 2, 0 ,K k i i 1 Electricity consumption optimization of production units M K d m ,k arg min Cm ,k k i i i 1 m 1 k 1 1 Power generation optimization of self generation power plant K pm ,k arg min Ck d m ,k i i i k 1 Update price by optimization direction ki 1 k ki 1 No Normalization ki i ki Convergence? Yes Output the final Price Figure 6-6 the flow chart of gradient information based price formation mechanism This mechanism has several advantages: a) It fully uses the historical information, the iteration process of internal price reach convergence rapidly; b) The final internal price is enough strong to be used as a dynamic price signal. c) The final internal price has the particular of both RTP and TOU tariff. It’s easy for practical use. And it also has several disadvantages: a) It has a little special requirement to meet for initial price. And all the features mentioned above will be analyzed and explained in case study shown in next chapter. 6.3.3 Heuristic Method Based Price Formation Mechanism RTP tariff mechanism 2 is a kind of heuristic method based price formation mechanism. And it has the iteration base on predefined price periods and centralized power optimization of end users. During iteration, the mechanism gives an increment to the price of the last iteration based on predefined price periods, which made by the prior knowledge. When the reception electricity is positive, it represents the enterprise is buying 24 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid electricity from utility. Then we give a increment of price like 1 kbuy kgene . The increment encourages users to reduce their electricity consumption, when kbuy kgene is positive, and to increase their electricity consumption, when the difference is negative. The mechanism combines the cost of buying electricity and generation for comparison ingeniously. It stands for smaller increment that the price for generation is closed to the price for buying electricity, which conforms to reality. When the reception electricity is negative, similar conclusion can be gained. The increment function is chosen as follow: f Qk i 1 kbuy kgene , if Qki 0 i 2 kbuy kgene , if Qk 0 i sell gene 3 k k , if Qk 0 (6-12) The steps of the price formation mechanism are as follows: Step 1: the energy management section offers a initial price as an internal price at will, k 1, 2, 0 k , K , set i=1; Step 2: under the i-th internal price, all production units optimize their electricity consumption, to minimize their total power consumption cost, and get the d m ,k , i M K d mi,k arg min C m i,k k i 1 ; (6-13) m 1 k 1 Step 3: the self generation power plant does the centralized power generation optimization, and get the optimal units output, K pm i,k arg min Cki d mi,k , (6-14) k 1 Step 4: using the internal price generated by the last iteration, and predefined price periods, compute the internal price k for next iteration, and the price i formulation is ki 1 k ki 1 M N 1 k f Qki , Qki d mi,k pmi ,k i m 1 n 1 f Qki (6-15) 1 kbuy kgene , if Qki 0 2 kbuy kgene , if Qki 0 i sell gene 3 k k , if Qk 0 25 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Step 5: realizing the normalization through k k , to make sure that the i i i total internal settlement cost is consistent before and after the Step 4. K i C i k k 1 M 1 1 f Qki ki 1 d mi,k i k 1 m 1 K (6-16) Step 6: if the total cost and price in this iteration doesn’t satisfy the condition of convergence: K i K i 1 K i Ck Ck / Ck C k 1 k 1 k 1 ; i i 1 max k k / i k (6-17) then i=i+1, go to step 2. Step 7: output and release the internal price k as a dynamic price signal in i microgrid. And the flow chart of this gradient information based price formation mechanism is shown in Figure 6-7. 26 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Energy management section offers initial price and set i=1 k 1, 2, 0 ,K k i i 1 Electricity consumption optimization of production units M K d m ,k arg min Cm ,k k i i i 1 m 1 k 1 1 Power generation optimization of self generation power plant K pm ,k arg min Ck d m ,k i i i k 1 Update price by optimization direction ki 1 k ki 1 No Normalization ki i ki Convergence? Yes Output the final Price Figure 6-7 a) b) c) d) the flow chart of heuristic method based price formation mechanism This mechanism has several advantages: The increment function intuitively reflect the guidance function of price signals. And those parameters can be adjusted for changing the strength of function; It fully uses the historical information, the iteration process of internal price reach convergence rapidly; The final internal price is enough strong to be used as a dynamic price signal. The final internal price has the particular of both RTP and TOU tariff. It’s easy for practical use. And it also has several disadvantages: a) It has a little special requirement to meet for initial price. And all the features mentioned above will be analyzed and explained in case study shown in next chapter.Equation Chapter (Next) Section 1 7 Test Result of Typical Cases and the Analysis 7.1 Cases Introduction 27 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises In order to observe and analyze the property of those mechanisms, two typical cases have been studied. Case 1 represents a typical small scale iron & steel manufactory enterprise. A period lasts 10 minutes and there are 144 periods totally. Based on primarily research results on an iron and steel plant, the case contains 3 kinds of equipments, which is ladle refining furnace (LF furnace), blooming mill and finishing mill. And also contain a self generation power plant called combined cycle power plant (CCPP). Figure 7-1 is the schematic diagram of technological process. The equipments list and their parameters are shown in TABLE 7-1. There are three preconditions during optimization time horizon: unlimited fuel supply for power generation; total electricity consumption keeping constant; Electricity users belonging to typical batch type 2. Emission Other Use Gas Holder Oxygen Making COREX Furnace Production Process CCPP Steelmaking (LF Furnace) Medium Plate Production Electricity Flow Grid Gas Flow Figure 7-1 the schematic diagram of technological process TABLE 7-1 No. Equipment 1 LF furnace 2 3 4 5 equipments list and parameters for case 1 Type Parameters Batch production Power rating: 20000 (KW) (Type 2) Batch time:50 (min) Blooming Batch production Power rating: 20000(KW) mill (Type 2) Batch time: 60 (min) Finishing Batch production Power rating: 40000 (KW) mill (Type 2) Batch time: 90 (min) Continuous Maximum output: 169000 (KW) production Minimum output:85000(KW) (Type 3) Δ Output :50000(KW/period) CCPP Gas holder Storage (Type 2) Minimum :160000 (Nm3) Maximum :500000(Nm3) number 3 3 1 1 1 Tariff: There are three general levels of TOU: peak price, valley price and flat price. And the feed-in tariff which happened when enterprise sold electricity to utility is smaller than the valley price. We use the TOU tariff implemented by utility in Shanghai 28 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid district, shown in TABLE 7-2. TABLE 7-2 TOU Tariff Implemented by the Utility Time period (hours) Tariff (¥/KWh) Valley 0:00-6:00,22:00-00:00 0.219 Flat 6:00-8:00,11:00-13:00, 15:00-18:00, 21:00-22:00 0.56 Peak 8:00-11:00,13:00-15:00, 18:00-21:00 0.926 Feed-in tariff 0:00-24:00 0.2 7.2 Pricing Formation Mechanism Based on Real-time Cost The numerical testing result of pricing formation mechanism is shown below. Figure 7-2 is the figure of the practical daily cost for enterprise during iterations. Figure 7-3 is the internal price curve and external TOU tariff curve during a day, which shows the relationship between the internal dynamic price and TOU tariff implemented by utility. Figure 7-4 is the total load curve and units output curve during a day, which intuitively shows the effect of this mechanism on production and generation scheduling. Case 1: 6 1.124 x 10 1.122 Daily Total Cost (¥ ) 1.12 1.118 1.116 1.114 1.112 1.11 1.108 1.106 0 5 Figure 7-2 10 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 29 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 1 Initial Price Final Price TOU Tariff 0.9 Internal Price (¥ /KWh) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 Figure 7-3 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-4 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Base on the result, a brief analysis is given: a) It almost decreases the total energy consumption cost of the whole enterprise. Because all the production equipments belong to batch equipment, and their load level is discrete, the fluctuation during iterations is inevitable; b) The internal price curve can be divided to three segments, and the time segments division is same as TOU tariff. In off-peak periods, the better choice is buying electricity. Because the output is pn at the first segment, the internal price is 30 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid concussive and has the opposite trend with load; c) At the second segment, the units output follows the total load to avoid buying electricity in peak period. Therefore the internal cost price is equal to generation cost price which is constant. And the internal price in the second segment is higher than the first and third segments definitely d) The internal price signal is too weak, and the concussive price is hard for practical use; e) The multiplicity of problem exists in this pricing formation mechanism. The practical daily cost for enterprise during iterations presents a severe concussion in case 2 although finding out the optimal solution. 7.3 Pricing Formation Mechanism Based on Gradient Information The numerical testing result of pricing formation mechanism is shown below. Figure 7-5 is the figure of the practical daily cost for enterprise during iterations. Figure 7-6 is the internal price curve and external TOU tariff curve during a day, which shows the relationship between the internal dynamic price and TOU tariff implemented by utility. Figure 7-7 shows the cost price, which compute by the practical cost and total load. Figure 7-8 is the total load curve and units output curve during a day, which intuitively shows the effect of this mechanism on production and generation scheduling. Case 1: 6 1.124 x 10 1.122 Daily Total Cost (¥ ) 1.12 1.118 1.116 1.114 1.112 1.11 1.108 1.106 0 5 Figure 7-5 10 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 31 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 1 Initial Price Final Price TOU Tariff 0.9 Internal Price (¥ /KWh) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 Figure 7-6 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 1 Initial Price Final Price TOU Tariff 0.9 Cost Price (¥ ) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 Figure 7-7 32 20 40 60 80 Time (10mins) 100 120 140 the cost price curve and external TOU tariff curve during a day Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-8 a) b) c) d) 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Base on the result, a brief analysis is given: The mechanism decreases the total energy consumption cost of the whole enterprise. Because it fully uses the historical information, the iteration process of internal price reach almost convergence rapidly, and the fluctuation during iterations is avoided in case 1; The trend of internal price is the same as external TOU tariff, and also can be treated as a like TOU tariff, which is a dynamic price changed among days. The internal price is enough strong to be a signal, and easy for practical use because of small ripple. The units output follows the total load to avoid buying electricity in peak period, which conform to reality. Therefore the internal cost price is equal to generation cost price which is constant shown in Figure 7-7. The mechanism realizing the normalization to make sure that the total internal settlement cost is consistent during iteration. So the cost price curve computed by the mechanism is the same as the curve computed by real-time cost based price formation mechanism, which is stand for the optimal solution. 7.4 Pricing Formation Mechanism Based on Heuristic Information The numerical testing result of pricing formation mechanism is shown below. Figure 7-9 is the figure of the practical daily cost for enterprise during iterations. Figure 7-10 is the internal price curve and external TOU tariff curve during a day, which shows the relationship between the internal dynamic price and TOU tariff implemented by utility. Figure 7-11 shows the cost price, which compute by the practical cost and total load. Figure 7-12 is the total load curve and units output curve 33 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises during a day, which intuitively shows the effect of this mechanism on production and generation scheduling. Case 1: 6 1.124 x 10 1.122 Daily Total Cost (¥ ) 1.12 1.118 1.116 1.114 1.112 1.11 1.108 1.106 0 5 10 Figure 7-9 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-10 34 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 1 Initial Price Final Price TOU Tariff 0.9 Cost Price (¥ ) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 Figure 7-11 20 40 60 80 Time (10mins) 100 120 140 the cost price curve and external TOU tariff curve during a day 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-12 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Base on the result, a brief analysis is given: a) The mechanism decreases the total energy consumption cost of the whole enterprise. Because it fully uses the historical information and heuristic information, the iteration process of internal price reach convergence rapidly, and the fluctuation during iterations is avoided; b) The trend of internal price is the same as external TOU tariff, and also can be 35 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises treated as a like TOU tariff, which is a dynamic price changed among days. The internal price is enough strong to be a signal, and easy for practical use because of small ripple. c) The units output follows the total load to avoid buying electricity in peak period, which conform to reality. Therefore the internal cost price is equal to generation cost price which is constant shown in Figure 7-11. d) The mechanism realizing the normalization to make sure that the total internal settlement cost is consistent during iteration. So the cost price curve computed by the mechanism is the same as the curve computed by real-time cost based price formation mechanism, which is stand for the optimal solution. The convergence costs of the three price formation mechanism are shown in TABLE 7-3. In this case, all the three mechanism have found the optimal solution. The item called net cost means the cost of buying electricity from utility. TABLE 7-3 the convergence costs of the three price formation mechanism Cost Item Cost Method Gradient Method Heuristic Method Net Cost 50071 50071 50071 Generation Cost 1057333 1057333 1057333 Total Cost 1107404 1107404 1107404 7.5 Expended Test and the Analysis In order to verify the influence of different parameters values, the expended situations are tested. The expended test includes changing the number of batch, changing the range of units output, 6 tests totally. 7.5.1 Change the Batch Number: Numbers of batch represent the density of production. In this expended test, the work intensity of equipments has been increased for observation. The heuristic method based price formation mechanism is used for the tests. a) Based on TABLE 7-1, change the batch number of LF furnace, blooming mill, and finishing mill from 3, 3, 1 to 5, 5 and 2. 36 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 6 1.175 x 10 Daily Total Cost (¥ ) 1.17 1.165 1.16 1.155 1.15 0 5 10 Figure 7-13 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-14 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 37 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 180 Initial Load Final Load Final Output 170 160 Power Level (MW) 150 140 130 120 110 100 90 80 0 20 Figure 7-15 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Although the daily total cost rises because of the increase of batch number, the convergence of the mechanism doesn’t change. And the trend of internal price is also same as external TOU tariff in Figure 7-14. And in Figure 7-15, we can easily find out the effect of production scheduling. The production activities are allocated to the off-peak hours as possible, and the units output track the load well to minimize the total cost. b) Further, change the batch number of LF furnace, blooming mill, and finishing mill from 3, 3, 1 to 7, 7 and 3. 38 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 6 1.1513 x 10 Daily Total Cost (¥ ) 1.1513 1.1513 1.1513 1.1513 1.1513 0 5 10 Figure 7-16 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-17 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 39 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-18 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Although the daily total cost rises because of the increase of batch number, the convergence of the mechanism doesn’t change. And the trend of internal price is also same as external TOU tariff in Figure 7-17. And in Figure 7-18, we can easily find out the effect of production scheduling. The production activities are arranged to the off-peak hours as possible, and the units output track the load well to minimize the total cost. 7.5.2 Analysis of Cost-saving For the proposed mechanism, decrease the cost of the whole enterprise is one of the most important tasks. In the numerical test, we found the ability of cost-saving is related to the production intensity. In this section, we discuss the capacity of saving energy consumption cost in different production intensity. In this expend test, we have three illustrations. a) The test shows us the capacity of saving cost, and the capacity is measured by the different between only power generation optimization and also using the dynamic price mechanism for power consumption optimization. Without the power generation optimization, the capacity of saving cost is hard to be measured. And if the power generation optimization is not used neither, the index of percentage cost decrease mentioned below would be much larger. b) To measure the production intensity, we use a index denoted PI, which is showed as: 40 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid PI total time consumption of production 100 0 0 total adjustable time of production M Jm T m 1 j 1 J em, j sm , j m 1 j 1 M (7-1) m Jm 100 0 0 which means the ratio of time consumption of production and available time can be used for production. c) To measure the capacity of saving energy consumption cost, we use a index denoted PCD, which is showed as: the cost decrease when using the mechanism the cost of optimal solution C Coptimal og 100 0 0 Coptimal PCD where the Cog means the cost when only use the power generation optimization. And Coptimal is the cost of optimal solution obtained by the proposed mechanism. When we increase the number of batch in different equipments, the PI index will increase, and the production intensity also increasing. The following table and shows the capacity of saving energy consumption cost in different production intensity. And the PI index is an ascending sequence. TABLE 7-4 the capacity of saving cost in different production intensity Number of batch in equipments No.2 1 No.3 1 PI Cog Coptimal PCD 1 No.1 1 9.90% 1.09E+06 1.08E+06 0.65% 2 2 1 1 14.85% 1.10E+06 1.09E+06 1.54% 3 2 2 1 17.33% 1.11E+06 1.09E+06 1.14% 4 3 2 1 22.28% 1.11E+06 1.10E+06 1.28% 5 3 3 1 24.75% 1.12E+06 1.11E+06 1.37% 6 4 3 1 29.70% 1.13E+06 1.12E+06 1.37% 7 4 4 2 34.65% 1.15E+06 1.13E+06 1.35% 8 4 5 2 37.13% 1.16E+06 1.14E+06 2.02% 9 5 5 2 42.08% 1.17E+06 1.15E+06 1.95% 10 5 6 2 44.55% 1.18E+06 1.16E+06 1.61% 11 6 6 2 49.50% 1.18E+06 1.17E+06 0.94% 12 6 7 3 54.46% 1.21E+06 1.19E+06 1.30% ID 41 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 7 7 Persentage Cost Decrease (PCD) 13 3 59.41% No solution 0.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% 0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% Production Intensity (PI) Figure 7-19 the capacity of saving cost in different production intensity When the production intensity is low, the mechanism optimizes the production process and decrease the energy cost. Comparing to the production consumption, the base load is still large, so the effect of deceasing the cost is not significant, which is reasonable. For this reason, when the PI is higher, more batches are scheduled, and the effect of mechanism is pretty good. No more space can be optimization, when the production intensity is extreme high, so the PCD start to decline. 7.5.3 Change the Range of Units Output The range of generation units output concern the ability of tracking the total load. In expended test, the range of unit output is changed for observation. Both heuristic method and gradient method based price formation mechanism are tested. TABLE 7-5 parameters and methods in tests Serial number Method Range(KW) original Both 85000-169000 a) Heuristic 105000-169000 b) Gradient 105000-169000 c) Heuristic 85000-109000 d) Gradient 85000-109000 e) Heuristic 65000-80000 f) Gradient 65000-80000 a) Heuristic method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units output from 85000-169000(KW) to 105000-169000(KW). 42 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 6 1.18 x 10 1.178 Daily Total Cost (¥ ) 1.176 1.174 1.172 1.17 1.168 1.166 1.164 0 5 10 Figure 7-20 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 2.5 Initial Price Final Price TOU Tariff Internal Price (¥ /KWh) 2 1.5 1 0.5 0 0 Figure 7-21 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 43 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-22 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day In both Figure 7-20 and Figure 7-21, there are fluctuations at some periods. Because all the production equipments belong to batch equipment, and their load level is discrete. Based on the analysis of Figure 7-22, for example, in period 50-100, units output is larger than the load, that makes the mechanism try to arrange the production activities here, and sell electricity at off-peak hours. It directly causes the rising cost. Therefore, at this situation in the mechanism, fluctuations can’t be avoided, and the mechanism needs to be improved; b) The gradient method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units’ output mill from 85000-169000(KW) to 105000-169000(KW). 44 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 6 1.18 x 10 1.178 Daily Total Cost (¥ ) 1.176 1.174 1.172 1.17 1.168 1.166 1.164 0 5 10 Figure 7-23 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.2 Initial Price Final Price TOU Tariff 1.1 Internal Price (¥ /KWh) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0 Figure 7-24 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 45 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-25 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day In both Figure 7-23 and Figure 7-24, there are fluctuations at some periods due to the same reason. And comparing to the heuristic method, the fluctuations are more serious. At this situation in the mechanism, fluctuations also can’t be avoided, and the mechanism needs to be improved; c) The heuristic method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units’ output mill from 85000-169000(KW) to 85000-109000(KW). 6 1.155 x 10 1.15 Daily Total Cost (¥ ) 1.145 1.14 1.135 1.13 1.125 1.12 1.115 1.11 0 5 Figure 7-26 46 10 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-27 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-28 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day Figure 7-26 proves that the mechanism is efficient when the upper limit of units output is lower than the peak load, which can be proved in Figure 7-28. And the internal price given by Figure 7-27 is proper and rational; d) The gradient method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units’ output mill from 85000-169000(KW) to 85000-109000(KW). 47 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 6 1.155 x 10 1.15 Daily Total Cost (¥ ) 1.145 1.14 1.135 1.13 1.125 1.12 1.115 1.11 0 5 10 Figure 7-29 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.8 Initial Price Final Price TOU Tariff 1.6 Internal Price (¥ /KWh) 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 Figure 7-30 48 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 160 Initial Load Final Load Final Output 150 Power Level (MW) 140 130 120 110 100 90 80 0 20 Figure 7-31 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day The figures show that the gradient information based mechanism is efficient when the upper limit of units output is lower than the peak load. The mechanism will be slightly worth than the heuristic method based mechanism in aspect of convergence. We also can find there is a pinnacle in Figure 7-29, which will be explained later; e) The heuristic method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units’ output mill from 85000-169000(KW) to 65000-80000(KW). 49 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 6 1.23 x 10 Daily Total Cost (¥ ) 1.22 1.21 1.2 1.19 1.18 1.17 0 5 10 Figure 7-32 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-33 50 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 160 Initial Load Final Load Final Output 150 140 Power Level (MW) 130 120 110 100 90 80 70 60 0 20 Figure 7-34 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day In this test, the generation units output can’t meet the requirement of the load of the whole enterprise, which can be proved by Figure 7-34. The test result shows that the effect in reducing total cost is excellent. And the internal price given by the mechanism in Figure 7-33 is proper and rational; The gradient method based price formation mechanism. Based on TABLE 7-1, change the range of CCPP units’ output mill from 85000-169000(KW) to 65000-80000(KW). 6 1.23 x 10 1.22 Daily Total Cost (¥ ) f) 1.21 1.2 1.19 1.18 1.17 0 5 Figure 7-35 10 15 20 25 30 Number of Iteration 35 40 45 50 the practical daily cost for enterprise during iterations 51 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises 1.4 Initial Price Final Price TOU Tariff 1.2 Internal Price (¥ /KWh) 1 0.8 0.6 0.4 0.2 0 0 Figure 7-36 20 40 60 80 Time (10mins) 100 120 140 the internal price curve and external TOU tariff curve during a day 160 Initial Load Final Load Final Output 150 140 Power Level (MW) 130 120 110 100 90 80 70 60 0 20 Figure 7-37 40 60 80 Time (10mins) 100 120 140 the total load curve and units output curve during a day The figures show that the gradient information based mechanism is efficient when the generation units output can’t meet the requirement of the total load. The mechanism will be slightly worth than the heuristic method based mechanism in aspect of convergence. There are pinnacles in Figure 7-29, Figure 7-32, et al. There are two figures below which show the total load curve under the same parameters, such as the same internal price. The figure left shows the load curve of the pinnacle during iterations, and the 52 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid right shows the load curve in a single iteration under the same parameters. The test figure below tells us that the price formation mechanism doesn’t cause pinnacles. The reason may be the ILOG CPLEX or electricity consumption optimization model of production units, which still need further research.Equation Chapter (Next) Section 1 15 Load in Iterations Load in a Single Iteration 14 Power Level (MW) 13 12 11 10 9 8 0 20 Figure 7-38 40 60 80 Time (10mins) 100 120 140 load curve in different iteration situation 8 Conclusion In this project, the electricity dynamic pricing problem in microgrid has been researched. Based on the analysis of the problem, an electricity pricing mechanism is proposed. The purpose and intent of the price mechanism includes two aspects: for the cost benefit, the mechanism tries to reduce the total electricity cost of microgrid by encouraging the end users to use less energy during peak hours, or to move the time of electricity use to off-peak hours. The optimal power strategy is also considered; at the same time, considering the usability, the price mechanism should be easy to implement, and be simple for operators. There are several key components in the price mechanism: a) Power consumption models for end users: different end users have different load control characteristics. b) Power generation models for self generation power plant: different power generator units have different output control characteristics and different generating cost. c) Iterative or game mechanism for reducing electricity cost: the iterative mechanism is convergent. 53 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises d) The division of time interval for price mechanism: The division of time interval is related with tariff implemented by the utility and generating cost. Though the analysis of real-time cost, the power consumption and generation optimization model is been built. Then the optimal power generation strategy has been made and several pricing formation mechanism has been designed. Based on the analysis of numerical test, the dynamic price mechanism demonstrates following points: a) With the optimal power generation strategy, final price is able to lead end user to shift load from peak hours to off-peak hours. Thus, a cost saving of about 1.5% is feasible comparing to that of initial load. b) Total income of energy management section charging to the end users with final price is equal to the total cost. c) With the optimal power generation strategy, three dynamic price mechanisms are able to reduce electricity cost of EIE microgrid. d) The RTP mechanism with heuristic rule is the best one based on current research. e) The ideal dynamic price electricity price curve is dynamic changing according to tariff implemented by the utility and self generation cost, f) The dynamic price electricity price is a specific real time pricing. In addition, we can conclude the applicable scope of the mechanism based on the analysis. That is the microgrid with adjustable load and self generation power plant, and the power units have good power generation performance. And the mechanism has a better performance when the following conditions are satisfied: a) Microgrid with adjustable load and self generation power plant, b) Power consumption has a larger adjustment range. c) Power units have good power generation performance, the installed capacity is close to the total load and the fuels for power generation are not limited. d) Relationship among generating cost and tariff implemented by the utility: Situation one: koff-peak kgene , kgene ksell , kpeak kgene And, when the situation is as follows, the dynamic price mechanism is no need to implement: a) Power outputs have good power generation performance, the installed capacity is close to the total load, the fuels for power generation are not limited. b) Relationship among generating cost and tariff implemented by the utility belong to situations as bellow: Situation two: koff-peak kgene ksell , kpeak kgene Situation three: koff-peak kgene , kgene ksell , kpeak kgene At last, when the power consumption has little adjustment range, the price mechanism is not applicable to implement.Equation Chapter (Next) Section 1 54 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid 9 The Description of General Dynamic Pricing Mechanism in Microgrid Base on the analysis of dynamic pricing mechanism in EIE microgrid, we can give a description of general dynamic pricing in microgrid. In traditional power supply pattern shown in Figure 1-1, the users hand in the charge of electricity according to the tariff given by utility. Then the electricity cost of energy management section (EMS) is zero. It can be expressed as follow: M Cm,k kbuy d m,k , Cktotal Cm ,k , Ckpay Cktotal m 1 gain k C C pay k C total k (9-1) 0 Transition to microgrid is shown in Figure 1-3. Both the utility and self-contained power supply form the multi-power supply pattern. All users hand in the charge of electricity to EMS in microgrid according to the internal tariff. And at the same time, EMS purchases the internal generation according to the internal tariff. The cost is expressed as follow. user gene gene Cmuser , k m , k d m , k , Cn , k n , k pn , k M N m 1 n 1 gene Ckuser Cmuser Cngene , k , Ck ,k (9-2) At this time, the power exchange cost (or benefit) between the utility and microgrid is undertaken by the EMS. The electricity cost of EMS is expressed as follow. Ckgate kbuyQk , if Qk 0 0 , if Qk 0 sell k Qk , if Qk 0 M N m 1 n 1 Qk d m,k pm ,k (9-3) Ckgain Ckuser Ckgate Ckgene Different tariff in microgrid brings different net income for EMS. There are three situations shown in TABLE 9-1 and TABLE 9-2. In situation 1, the internal price is the tariff implemented by the utility which is a single time period. And the internal power demand/generation will bring income for the EMS. In situation 2 and 3, the internal price TOU tariff implemented by the utility which has peak hour and off-peak hour. Different tariff will adjust the power demand/generation, which means different income for EMS. 55 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises TABLE 9-1 the items value in situation 1 Items Situation 1 Power demand 1000KWh Power generation 200KWh Net power demand 800KWh Tariff implemented by the utility 1$/KWh Net electricity cost 800$ Internal power demand price 1$/KWh Internal power generation price 0.3$/KWh Internal electricity cost 1000$ Internal generation cost 60$ Net income for EMS 1000-60-800=140 $ TABLE 9-2 Items the items value in situation 2 and 3 Situation 2 Situation 3 Peak hour Off-peak hour Peak hour Off-peak hour Buying tariff 1 $/KWh 0.5 $/KWh 1 $/KWh 0.5 $/KWh Feed in tariff 0.3 $/KWh 0.1 $/KWh 0.3 $/KWh 0.1 $/KWh Power demand 1000 KWh 1000 KWh 700 KWh 1300 KWh Power generation 200 KWh 200 KWh 300 KWh 100 KWh Net electricity cost 800 $ 400 $ 400 $ 600 $ Internal electricity cost 1000 $ 500 $ 700 $ 650 $ Internal generation cost 60 $ 20 $ 90 $ 10 $ Internal cost 940 $ 480 $ 610 $ 640 $ Net income for EMS 140 $ 80 $ 210 $ 40 $ Total net income for EMS 220 $ 250 $ In this situation, the users pay the charge according to the internal dynamic electricity price, and arrange the electricity for reducing electricity cost. Meanwhile, the generators receive repayment according to the internal dynamic generation price, and arrange the generation in each period for increasing income. It brings a problem. That is how to distribute the benefit for win-win, which can be assigned to game. And a dynamic pricing based solution is proposed. The flow chart of this generalized game mechanism is shown in Figure 9-1. 56 Dynamic Pricing Model and Algorithm in Energy Intensive Enterprise Microgrid Initial setting Power users K d arg min C m ,k user m ,k k 1 muser ,k No Power generators K p arg max C n ,k k 1 gene n ,k d m ,k , ngene ,k Energy management section gene n ,k , arg max C K user m ,k k 1 gain k d m,k , pm,k , kbuy , ksell Equilibrium? Yes End Figure 9-1 the flow chart of generalized game mechanism Different types of microgrid need different price mechanisms. In view of classification of price mechanism, the microgrid can be classified according to: a) Demand response of power users: centralized scheduling or decentralized decision making; b) Demand response of power generators: centralized scheduling or decentralized decision making; c) Integration degree of power users and power generators; d) The benefit-based relationships among power generators, power users and energy management section 57 Dynamic Pricing in Microgrid for Energy Intensive Industrial Enterprises Reference [1] Triki Chefi, Violi Antonio. “Dynamic pricing of electricity in retail markets,” A Quarterly Journal of Operations Research, vol. 7, no. 1 ,pp: 21-36, Mar 2009; [2] George B. Dantzig and Mukund N. Thapa. Linear programming 1: Introduction. Springer-Verlag, 1997. [3] George B. Dantzig and Mukund N. Thapa. Linear Programming 2: Theory and Extensions. Springer-Verlag, 2003. 58