Multi-energy Complementary Power System Economic Dispatch Considering Carbon Emission and Regional User Coordination

The integration of multi-energy complementarity and source-grid-load-storage is an important initiative to promote energy transformation and the high-quality development of power systems. This paper proposes a two-tier day-ahead multi-energy complementary power system economic dispatch model from the perspective of clean and low-carbon, taking into account carbon emissions and multi-regional user synergy. The model optimises the production and purchase of electricity through different decisions between the upper generation side and the lower user side, which minimises the energy production cost in the upper tier and the energy purchase cost in the lower tier, and achieves mutual benefits between the upper and lower tiers. This paper validates the algorithm and model with an improved IEEE 39-node system.


Introduction
"The 14th Five-Year Plan" is the key period and window for achieving the carbon peak, and building a new power system with new energy sources as the mainstay is a must for promoting energy transformation and achieving the "carbon peaking and carbon neutrality" target.
A large number of scholars at home and abroad have conducted relevant research from different perspectives on new power systems with a high proportion of new energy sources.In terms of new energy consumption, in [1], a stochastic optimization model for a hybrid wind-storage system participating in joint grid dispatch was developed by constructing different wind power scenarios with the objective of maximising returns.Liu et al. [2][3] [4] use the concept of Conditional Value at Risk (CVaR) in economics in different problems to measure the cost of risk loss caused by uncertainty on dispatch, which improves the phenomenon of wind and light abandonment.In terms of environmental benefits, Cheng et al. [5] incorporate a carbon tax into the demand response process based on the emission rates of coal-fired units, reducing the level of CO 2 emissions during dispatch.Wang et al. [6] incorporate carbon emission issues into the optimisation process through the MOP model.In terms of demand response, in [7], a top-down Hierarchical Optimization (HO) model was proposed to optimise the cost of energy purchase and improve the efficiency of energy use.Hu et al. [8] introduce a Heuristics-Based Demand Response (HDR) model for residential-level energy use dispatch, which helps to reduce the peak-to-valley difference in the grid.In general, with the development of new power systems, there is a need to coordinate and interconnect the energy production side and the consumer side to achieve mutual benefits.
To this end, this paper establishes a two-tier structured dispatch model with an upper generation side and a lower user side, as shown in Figure 1.The upper tier floating tariff model corrects the lower tier purchased tariff, which affects the lower tier user's energy purchase plan; the lower tier demand response model based on load transfer matrix (LTM) optimises the user's energy consumption plan, so that the lower tier user side smooths the load curve of the upper tier generation side.Through the continuous alternating iterations of transfer variables between the two tiers of the system, the production and purchase of electrical energy are optimised jointly, which improves the overall efficiency of the whole system operation.

Generation-side day-ahead dispatch model
The power generation side is a new energy-based multi-energy complementary power generation system, including thermal power, hydropower, wind and PV storage, and is on the upper level of the whole framework.In the upper model, this paper proposes an energy cost minimisation model: (1) where th C , h C , bat C and ren C are the costs of thermal power generation, hydroelectric power generation, energy storage and new energy respectively.In the day-ahead model, the number of dispatch periods T is taken as 24.
The upper tier model obeys unit constraints, system balance constraints, and lower tier power purchase constraints.

Upper-tier multi-energy complementary generation side model
In Equation (1), the cost of thermal units includes fuel cost, carbon emission cost and O&M cost; the cost of hydropower units is O&M cost; the cost of new energy includes O&M cost and risk loss cost including CVaR; the cost of energy storage includes O&M cost and aging cost.For reasons of space, this paper does not go into detail here.

Floating tariff model
This paper proposes a floating tariff model based on the fluctuation of nodal transfer carbon emissions and equivalent new energy, using the "base tariff + floating tariff" for the sale/purchase of electricity at each point in time.On the basis of the original time-sharing tariff, the tariff at nodes with low-carbon emissions or a high proportion of new energy equivalent output fluctuates downwards, and vice versa the tariff fluctuates upwards.The floating tariff model is as follows: q t q t q t q t r r where revised , q t r denotes the corrected tariff at moment t of the lower tier system connected to the upper tier node q; grid , q t r denotes the pre-correction tariff; This paper constructs a floating tariff correction factor carbon , q t σ based on the emission factors [10] of different thermal power units: where Γ denotes the set of grid-side power purchase nodes connected to the CCHP system; Φ denotes the set of nodes where the thermal power unit is located; carbon , q t acc denotes the share of the carbon emission level at power purchase node q at moment t in the total carbon emission of the system at the corresponding moment; Con j is the equivalent CO 2 conversion factor corresponding to the pollution level of the j th emission [11] ; , k j

G
is the pollutant emission factor indicating thermal power unit k for pollutant j; τ is the total number of load nodes in the upper system.Similarly, the floating tariff correction factor ren , q t σ is shown below: where Ω denotes the set of nodes where the wind and PV plants are located; avg , q k P denotes the transferred generation from the predicted average output of new energy units at node k to the power purchase node q.The lower user-side model obeys the individual unit constraints in the CCHP system, demand response constraint, time-of-use tariff adjustment constraint, power purchase constraint, and heat and cold power balance constraint.
The user-side electricity purchase cost model is shown below: where grid , i t P is the amount of electricity purchased/sold by users in lower region i.This paper proposes a method for defining the load transfer matrix, which quantitatively describes the load transfer behaviour of flexible loads during demand response.
, 11 , where the non-diagonal element Assuming that the electrical load in area i at time t is

T i t i th i h h
In summary, the demand response incentive model proposed in this paper is shown below: ( ) where 1 μ , 2 μ and 3 μ are incentive factors; EV , i t P and EV , i t P ′ denote the EV load in region i at time t and the load after the demand response, respectively.The specific algorithm is similar to that for the electrical load and will not be repeated here.

Introduction of a simulation example
In this section of the simulation analysis, the upper generation side is exemplified by an improved IEEE-39 system, containing 10 thermal units, 3 hydro units, and a wind farm and PV plant equipped with energy storage and connected to three regional CCHP systems.The lower user side is analysed with three typical district CCHP systems containing commercial, residential and industrial areas.This section will simulate and analyse three examples under different factors.Case 1 considers the full range of factors.Based on Case 1, Case 2 does not take into account the tariff regulation effect.Based on Case 1, Case 3 does not take into account the demand response of the lower user side.

Analysis of simulation results
The revised tariffs and tariff adjustments for each region in the lower tier are shown in Figures 2(a) and 2(b) respectively.In Figure 2(a), the fluctuation trend of the revised tariff in each region follows the change of the net load curve in real time, indirectly reflecting the fluctuation of the new energy output of the upper system.On the other hand, Figure 2(a) shows that the revised tariffs for the commercial and industrial areas are higher than those for the residential areas, corresponding to a higher level of transferred carbon emissions, reflecting the idea of "high carbon, high price, low carbon, low price".The change in user-side electricity purchases in the lower tier leads to a change in the system load curve on the upper tier generation side, as shown in Figure 4, where 1.31% of the upper tier system load curve is reallocated.In the original load curve of the upper tier system, 9:00 to 12:00 and 17:00 to 20:00 correspond to the two load peaks, while after considering the LTM-based demand response in the lower tier, the load is reallocated for this period, with 37.74% shifted to 14:00 to 16:00 and 62.26% shifted to 21:00 to 6:00.Table 1 shows the optimisation results for the upper generation side under different calculations.Compared to Case 2, the total cost of Case 1 is reduced by $ 6,633, with the cost of thermal power reduced by 0.22%, the cost of new energy reduced by 1.14% and the cost of energy storage reduced by 10.54%.Compared to Case 3, the total cost of Case 1 was reduced by $ 9,354, with the cost of thermal power reduced by 0.69%, the cost of new energy reduced by 1.51% and the cost of energy storage increased by 2.56%.Table 2 shows the optimisation results for the lower user side in different examples.Compared to Case 2, Case 1 takes into account the tariff regulation effect and increases the peak-valley tariff difference, which leads to a large amount of transferable load shifting to lower tariff periods and increases the incentive for users to participate in demand response, reducing electricity purchase cost by 1.50% and increasing the demand response incentive by 5.94%.Compared to Case 3, Case 1 takes into account the LTM-based demand response model, resulting in a 17.11% reduction in total cost.

Conclusions
This paper establishes a two-tier day-ahead joint dispatch model between the generation side and the consumer side, and interconnects the energy production side and the consumer side through a floating tariff model and a demand response model to achieve mutual benefits between the upper and lower levels.

Figure 1 .
Figure 1.Two-tier system day-ahead dispatch structure diagram reference [12]; grid C is purchasing electricity cost on the user side; DR C is demand response incentive cost.The models for grid C and DR C are designed as below.
denotes the share of load shifted from moment h to moment t in the th i region; the diagonal element , i tt sft denotes the share of load not participating in the demand response at moment t in the th i region.

Figure 2 .Figure 3 .
Figure 2. Modified tariff for floating tariff modelThe general electricity load and EV load before and after the demand response in each lower level region are shown in Figure3(a) and (b) respectively.As can be seen from Figure3, under the LTMbased demand response, the load decreases during peak load periods and increases during other periods, quantitatively describing the load shift.

Figure 4 .
Figure 4. Upper tier load curve obtained after user-side demand response of Case 1

Table 1 .
Comparison of optimization results on the upper generation side

Table 2 .
Comparison of optimization results on the lower user side