7.1 Concept of Multi-energy Management

7.1.1 Motivation and Background

Generally, all the energy systems are “multi-energy systems” in the sense that multiple energy sectors interact at different levels. For example in conventional power systems, the coal or gas used for generating electricity should be transported to each power plant, and this process implies the couplings between fossil energy and electrical energy. Another case is, the heating service by the combined heat-power plant also last for decades, and this process includes the coupling between heating energy and electrical energy. However, those energy couplings between different systems are conventionally weak compared with the relationship within a single energy system, and that is the main reason for the past studies of power system mostly only consider the electrical energy [1,2,3]. However, the interactions between different energy systems become tighter and more frequent recently, and this trend is about to continue in the future [4,5,6,7], such as the electric-gas energy system, and the coordinated heat-power system, or even the transportation-power system motivated by the transportation electrification. In this sense, conventional energy management for a single energy system may not be valid in the future, which drives the research of multi-energy management.

In literature, [8,9,10,11] focus on the coordination between the gas system and power systems [12,13,14,15]. Study the energy management methods for heat-power systems [16, 17]. Study the water-power systems and [18,19,20,21,22] investigate the coupling between the transportation system and power system by electric vehicles’ charging and discharging. The above research has brought a new perspective in energy system analysis, particularly in the light of reducing the economic and environmental burden of energy services. In summary, three benefits can be achieved by multi-energy management:

  1. a.

    Increasing or improving the energy efficiency of the entire system and the utilization of primary energy sources. The reason is the multi-energy system can use the energy at different levels. For example, the waste heat after generating electricity can be used for heating services and the energy efficiency of the entire system improves.

  2. b.

    Better deploying various energy resources at multiple system levels. For example, small-scale gas turbines can respond to volatile electricity market prices in a wind-rich energy system.

  3. c.

    Increasing the energy system flexibility by the coordinations between different energy systems. For example, scheduled charging/discharging of the electric vehicles acts as demand response tool for power system. Or the thermal storage tank can bring flexibility for combined power-heat plants.

Since the above three main advantages, the research on multi-energy management is essential for future energy systems. However, different energy systems have different administrators and quite distinct characteristics, and their coordinations are much more complex compared with the coordinations within a single energy system. Proper modeling methods and control strategies should be proposed to facilitate their operation.

7.1.2 Classification of Multi-energy Systems

The multi-energy systems can be classified by different perspectives, and there are mainly four perspectives to characterize the MES. The first is the spatial perspective. This perspective points out how MES can intend at different levels of aggregation in terms of components or even just conceptually. These levels go from buildings to district and finally to regions and even countries. This classification is shown in Fig. 7.1a.

Fig. 7.1
figure 1

Classifications for MES [4]

Full size image

The second perspective focuses on the provision of multiple services by optimally scheduling different energy systems, particularly at the supply levels. Such as the services provided by the MES, including electricity supply, water supply, heating service, EV charging services, gas filling services, and so on. This classification is shown in Fig. 7.1b.

The third perspective highlights how different types of fuels can be integrated together for providing optimal energy services, typically for economic or environmental targets. The fuel types range from classical fossil fuel, such as oil, coal and natural gas, to biomass fuels, and renewable energy. This classification is shown in Fig. 7.1c.

The fourth perspective discusses the coordinations between different energy systems, especially the coordination between different networks, such as the electrical network, gas network, district heating/cooling network, in terms of facilitating the development of multi-energy management methods and their interactions. This classification is shown in Fig. 7.1d.

Figure 7.1a classifies the MESs from the spatial perspective. An individual building exchanges energy by the transmission of electricity, heat, cooling, and natural gas. Then multiple buildings aggregate as a district, then multiple districts aggregate as a region and expand to a wider area. In this perspective, MESs can be classified as the building MES, district MES, region MES, and so on.

Figure 7.1b classifies the MESs from the service perspective. Generally, MES can provide multiple services to the customers, such as the electricity supply, heat and cooling power, and even some transport services, such as the charging/discharging of EV. In this perspective, MESs can be classified as combined electric-heat MES, combined electric-heat-cooling MES, and even electric-heat-water supply MES, since the water pump is coupled with the electric network by the electrical water pumps.

Figure 7.1c classifies the MES from the fuel perspective. For example, there exist many power sources in MES, such as power plants, boilers, gas turbines, and chillers. They may consume different types of fuels. Different power plants may consume coal, oil, or gas. A boiler may consume electricity or other fossil fuel, and a chiller may consume electricity or heat power. In this sense, the fuel type can also classify the MESs, such as the coal-gas MES, gas-hydrogen MES, or even ammonia MES since ammonia is a new type of carbon-free fuel [23].

Figure 7.1d classifies the MES from the network perspective since every “energy carrier” should be transmitted by a designed network. The electrical network includes power systems on multiple scales. Gas and oil are transported by pipelines or transportation flows. Heat and cooling power also have certain pipelines. Those different networks can have different topologies and operating strategies, which is the main motivation of this classification method. In this sense, the networks of MESs can be classified as combined electric-heat networks, combined electric-heat-cooling networks, and so on.

7.2 Future Multi-energy Maritime Grids

7.2.1 Multi-energy Nature of Maritime Grids

A sketch of MES is given in the former section to show the basic advantages and characteristics. In this section, the multi-energy nature of maritime grids will be analyzed to show their similarities and differences compared with conventional MESs, and Fig. 4.1 is re-drawn below as Fig. 7.2 as an illustration of future maritime grids. Two cases of maritime grids will be given after this illustration.

Fig. 7.2
figure 2

Illustrations of future maritime grids

Full size image

(1) Spatial perspective

From Fig. 7.2, maritime grids cover different spatial areas. For example, island microgrids cover an individual island, and the energy sources include offshore wind power, photovoltaic power, and underground cables. Seaport microgrids cover the harbor territory, and the energy sources include the offshore wind farm, land-based photovoltaic farm, oil pipelines, and the electricity supply from the harbor city. Other maritime grids include the drilling platforms and different types of ships. In summary, maritime grids have a very wide range on system scales, from the smallest to a ferry or a building and the biggest to a harbor city, which involves all the energy sources within a conventional MES. Different maritime grids are coupled tightly by energy connections, and current multi-microgrid coordination methods can be used in maritime grids to achieve better system characteristics.

(2) Service perspective

Figure 7.2 shows maritime grids can provide different services to customers, including the conventional services of electricity, heat, cooling in land-based MES, also including some types beyond current focuses, such as the logistic services, fuel transportation services. This is the primary difference between current studied MES (land-based MES) from the maritime grids. This is also a challenge for the research of maritime grids, since new energy models of those services should be formulated and integrated into the energy management model.

(3) Fuel perspective

Maritime grids also involve different types of fuels. In Fig. 7.2, the drilling platform can harvest crude oil or natural gas, and transport them to an island or the seaport. The industrial factory can refine crude oil into different types of fuels, such as gasoline, diesel, and so on. Those fuels may in reverse fill into the ships for sailing, into seaport for generation, and into the harbor city for daily lives. Besides, some novel fuels may also use in maritime grids, such as ammonia, methanol, and ethanol.

(4) Network perspective

Maritime grids also have different types of networks. Figure 7.2 shows some typical ones, (1) electrical networks in harbor city, seaport, industrial factory; (2) heat/cooling networks in harbor city, seaport, industrial factory; (3) fossil fuel networks between the ocean platforms and a seaport or an island; (4) electrical networks between offshore wind farms and a seaport or an island; (5) multi-energy network within an island; (6) transportation network by ships and vehicles. Those networks above are connected with multiple energy and information flows and may be more complex than conventional land-based MESs.

7.2.2 Multi-energy Cruise Ships

In Fig. 7.3, a typical topology of a multi-energy cruise ship is shown. Detailed illustrations can be depicted as follow. The load demands can be classified into three categories, the electric load, thermal load, and propulsion load. Among the three load demands, the propulsion load is to drive the cruise ship, which consists most of, usually more than 50% of the total load demand [24]. The propulsion load has a simple cubic relationship with the cruising speed, which is under the constraints of navigation distance [25]. The electric load in cruise ships includes the illumination, recreation equipment, movie theater, and so on. This type of load scales up to tens of megawatt [24], which is provided by the electric power bus, shown as the blue lines and arrows in Fig. 7.3. The thermal load in cruise ship includes the cooling and heat load, the swimming pool, and the cooking. This type of load also scales up to tens of megawatt [27], which is provided by the thermal power network, shown as the green line and arrows in Fig. 7.3. It also should be noted that in some cruise ships the cooling and heat loads are provided by the electricity. In this work, we will compare the introduced multi-energy technology with the single electric supply mode.

Fig. 7.3
figure 3

Topology of a multi-energy cruise ship

Full size image

As for the generation systems, to provide adequate electric and thermal loads for the overall cruise ship. There exist three types of generation systems, i.e., DG, CCHP, and PTC. The DGs make up the main part of the shipboard generation, which provides most of the electric power supply. The CCHP both provides the electric power and the thermal power and the PTC uses electricity to produce thermal power. To balance both the electric and thermal loads, the HES (electric and thermal energy storage) is integrated.

7.2.3 Multi-energy Seaport

We have illustrated the multi-energy seaport in Chap. 1. Here we re-draw Fig. 1.17 as Fig. 7.4to further show its multiple energy flows.

Fig. 7.4
figure 4

Multi-energy seaport microgrid

Full size image

Generally, the seaport is connected with the main grid and various renewable energy are integrated, i.e., seaport wind farms and PV farms in Fig. 7.4. All the port-side equipment, including the quay cranes, gantry cranes, transferring trunks, are electrically-driven. The seaport provides four types of services to the berthed-in ships and has four sub-systems for each type of services: (1) logistic service. The berth allocation and quay crane scheduling for loading/unloading cargo; (2) fuel transportation. Unloading or refilling fuel for the berthed-in ships; (3) cold-ironing. Providing electricity to the berthed-in ships and (4) refrigeration reefer for the cold-chain supply. The coordination between different sub-systems is shown in Fig. 7.5. Four sub-systems are communicating by the seaport control center and a distributed control strategy is employed in the seaport microgrid.

Fig. 7.5
figure 5

Coordination between different sub-systems in seaport microgrid

Full size image

7.3 General Model and Solving Method

7.3.1 Compact Form Model

From above, maritime grids involve different networks and provide multiple types of services by different types of fuels. In this sense, maritime grids have a significant characteristic, i.e., using the electric network as the backbone for energy management, and other different networks serve as the “load demand” of electric networks. For example, the heat/cooling networks couple with the electric network by CHP or electric boiler/chiller, and water supply network couple with the electric network by electric water pumps, and logistic network couple with the electric network by charging/discharging.

For this complex network, a general energy scheduling form can be shown as (7.1). Where \( f\left( x \right) \) is the objective function of the main network, generally the electric network, and \( x \) is the decision variable vector; \( g_{i} \left( {y_{i} } \right) \) is the objective function of the i-th network, and \( y_{i} \) is the decision variable vector of the i-th network; \( F\left( x \right) \) is the constraint set of the main network; \( G_{i} \left( {y_{i} } \right) \) is the constraint set of the i-th network; \( A_{i} \cdot x = H_{i} \left( {y_{i} } \right) \) is the coupling constraint set of power consumption of coupling equipment, such as water pump, CHP, and various logistic equipment.

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{x,y_{i} }} f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \\ {s.t. F\left( x \right) \ge 0,G_{i} \left( {y_{i} } \right) \ge 0} \\ {A_{i} \cdot x = H_{i} \left( {y_{i} } \right),x \in X,y_{i} \in Y_{i} } \\ \end{array} $$
(7.1)

7.3.2 A Decomposed Solving Method

This Chapter proposes a decomposed method to solve this type of problem, which is given by the following Theorem 7.1.

Theorem 1

The above formulation is equivalent to the following form.

$$ \begin{array}{*{20}c} {\mathop { \hbox{min} }\limits_{x} \left[ {f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} \mathop {inf}\limits_{{y_{i} ,\tau_{i} ,u_{i} }} \left( {g_{i} \left( {y_{i} } \right) + \tau_{i} \cdot G_{i} \left( {y_{i} } \right) + u_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot x} \right]} \right)} \right]} \\ {s.t. F\left( x \right) \ge 0,G_{i} \left( {y_{i} } \right) \ge 0} \\ {\begin{array}{*{20}c} {x \in X \cap V} \\ {\begin{array}{*{20}c} {V \equiv \mathop \cup \limits_{i = 1}^{n} \left\{ {x|\lambda_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot x} \right] = 0} \right\}} \\ {, where \lambda_{i} \ge 0 and\mathop \sum \limits_{i = 1}^{n} \lambda_{i} = 1} \\ \end{array} } \\ \end{array} } \\ \end{array} $$
(7.2)

where \( u_{i} \) is the optimal multiplier vector of the following optimization problem.

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{y_{i} }} \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \\ {s.t.G_{i} \left( {y_{i} } \right) \ge 0,A_{i} \cdot x = H_{i} \left( {y_{i} } \right),for\,all\,i} \\ \end{array} $$
(7.3)

Proof

(1) Problem (7.1) and (7.2) have the same feasible region.

(1.1) If \( \bar{x} \) be feasible for (7.1), then \( \bar{x} \) is feasible for (7.2).

Let \( \bar{x} \) be an arbitrary point in the feasible region of (7.1), then

$$ F\left( {\bar{x}} \right) \ge 0,A_{i} \cdot \bar{x} = H_{i} \left( {y_{i} } \right),for \forall i $$
(7.4)

It follows that (7.5) holds for all \( \lambda_{i} \).

$$ \lambda_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot \bar{x}} \right] = 0 $$
(7.5)

Then \( \bar{x} \in V \), and \( F\left( {\bar{x}} \right) \ge 0 \). \( \bar{x} \) is also feasible for (7.2).

(1.2) If \( \bar{x} \) be feasible for (7.2), then \( \bar{x} \) is feasible for (7.1).

Let \( \bar{x} \) be an arbitrary point for (7.2), then (7.5) holds at least for one i. \( F\left( {\bar{x}} \right) \ge 0 \) is satisfied all the same, then (7.6) holds.

$$ \eta \cdot F\left( {\bar{x}} \right) + \lambda_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot \bar{x}} \right] \ge 0 $$
(7.6)

It follows that

$$ \mathop {\text{Inf}}\limits_{\eta \ge 0} \left\{ {\eta \cdot F\left( {\bar{x}} \right) + \lambda_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot \bar{x}} \right] \ge 0 } \right\} $$
(7.7)

Since \( \eta = 0 \) is allowed in (7.7). Now, (7.7) is the dual of the following optimization problem.

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{y_{i} \in Y_{i} }} 0^{T} \cdot y_{i} } \\ {s.t.F\left( x \right) \ge 0,H_{i} \left( {y_{i} } \right) = A_{i} \cdot \bar{x}} \\ \end{array} $$
(7.8)

Obviously, (7.8) is feasible and has the optimal value of 0, hence, \( \bar{x} \) is feasible for (7.1).

(2) The objective function

Since \( u_{i} \) is the optimal multiplier vector of (7.3), then (7.9) holds.

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{y_{i} }} \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \\ { = inf\left\{ {\mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right) + \mathop \sum \limits_{i = 1}^{n} \tau_{i} \cdot G_{i} \left( {y_{i} } \right) + \mathop \sum \limits_{i = 1}^{u} u_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot x} \right]} \right\}} \\ \end{array} $$
(7.9)

In this sense,

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{x,y_{i} }} [f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)]} \\ {\mathop { \hbox{min} }\limits_{x} \left[ {f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} \mathop {inf}\limits_{{y_{i} ,\tau_{i} ,u_{i} }} \left( {g_{i} \left( {y_{i} } \right) + \tau_{i} \cdot G_{i} \left( {y_{i} } \right) + u_{i} \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot x} \right]} \right)} \right]} \\ \end{array} $$
(7.10)

From above, (7.1) and (7.2) are equivalent, then the solution process is given below.

Solution process: From (7.2), the original problem can be solved in a two-step process. It should be noted that, \( g_{i} \left( {y_{i} } \right) + \tau_{i} \cdot G_{i} \left( {y_{i} } \right) \) is a constant when minimizing \( x \), so it is eliminated for simplification.

Step 1: Given a feasible \( \bar{x} \), solve (7.11) for \( y_{i}^{*} \) and \( u_{i} \).

$$ \begin{array}{*{20}c} {\mathop {\hbox{min} }\limits_{{y_{i} }} \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \\ {s.t.G_{i} \left( {y_{i} } \right) \ge 0,A_{i} \cdot \bar{x} = H_{i} \left( {y_{i} } \right),\,for\,all\,i} \\ \end{array} $$
(7.11)

It should be noted that, there are no coupling between different networks. So (7.11) can be solved in parallel.

Step 2: Solve (7.12) for \( x. \)

$$ \begin{array}{*{20}c} {\mathop { \hbox{min} }\limits_{x} \left[ {f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} \left( {u_{i} \cdot \left[ {H_{i} \left( {y_{i}^{*} } \right) - A_{i} \cdot x} \right]} \right)} \right]} \\ {s.t.F\left( x \right) \ge 0,A_{i} \cdot x = H_{i} \left( {y_{i}^{*} } \right),\,for\,all\,i } \\ \end{array} $$
(7.12)

Then check the convergence characteristic, if yes, terminates and if not, return to Step 1 and update \( \bar{x} \). The algorithm convergence is given below.

Algorithm convergence

It is proved that the proposed method has finite \( \varepsilon \)-convergence characteristic.

Theorem 2

Assume \( X \) and \( V \) are both compact set, \( f \), \( g \), \( F \), \( G_{i} \) and \( H_{i} \) are continuous. The set \( UT\left( x \right) \) of the optimal multiplier vector for (7.3) is non-empty for all \( x \) in \( X \) and uniformly bounded. Then, for any given \( \varepsilon > 0 \), the proposed procedure terminates in a finite number of steps.

Proof

For simplification, we define (7.13).

$$ L\left( {x,\tau ,u} \right) = f\left( x \right) + \mathop \sum \limits_{i = 1}^{n} \left( {g_{i} \left( {y_{i} } \right) + \tau \cdot G_{i} \left( {y_{i} } \right) + u \cdot \left[ {H_{i} \left( {y_{i} } \right) - A_{i} \cdot x} \right]} \right) $$
(7.13)

For any sequence \( L\left( {x^{v} ,\tau^{v} ,u^{v} } \right),x^{v} \) of the optimal solution of (7.2). Firstly, the optimal multipliers sequence \( \tau^{v} \), \( u^{v} \) will converges to a point noted as (\( \bar{\tau },\bar{u} \)), since the uniformly bounded assumption of \( UT\left( x \right) \). Additionally, \( x^{v} \) will converge to a point denoted as \( \bar{x} \) since the compactness of \( X. \)

At last, since \( L\left( {x^{v} ,\tau^{v} ,u^{v} } \right) \) is a non-increasing sequence and bounded below, there exists at least one sub-sequence of \( L\left( {x^{v} ,\tau^{v} ,u^{v} } \right),x^{v} \) which converges to a point, we noted it as \( L\left( {\bar{x},\bar{\tau },\bar{u}} \right),\bar{x} \).

Since the weak duality, (\( \bar{\tau },\bar{u} \)) is the optimal multiplier for \( \bar{x} \) and (7.14) holds.

$$ L\left( {\bar{x},\bar{\tau },\bar{u}} \right) = \mathop {\text{Inf}}\limits_{{y_{i} }} \left( {f\left( {\bar{x}} \right) + \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \right) $$
(7.14)

Then, for any given \( \varepsilon > 0 \), there should be finite \( v \) to make (7.15) hold.

$$ L\left( {\bar{x},\bar{\tau },\bar{u}} \right) \le L\left( {x^{v} ,\tau^{v} ,u^{v} } \right) \le \mathop {\text{Inf}}\limits_{{y_{i} }} \left( {f\left( {\bar{x}} \right) + \mathop \sum \limits_{i = 1}^{n} g_{i} \left( {y_{i} } \right)} \right) + \varepsilon $$
(7.15)

Then the proposed method should converge in finite steps.

7.4 Typical Problems

7.4.1 Multi-energy Management for Cruise Ships

This section uses the cruise ship in Fig. 7.3 as the test case to show the effects of energy management. For a more economic and environmental operation of the cruise ship, the shipboard energy management system will optimally dispatch the outputs of the DG, CCHP, PTC, and HES to fulfill the propulsion, onboard electric, and thermal loads. However, in practice, those control variables are not on the same time-scale. During the navigation, the ship will constantly cruise and the speed cannot be regulated rapidly [25], and the onboard facilities for tourists also should keep working till night. This makes the propulsion and electric loads should be fulfilled in a long-term horizon (every hour in this work). Besides, the thermal load should be satisfied in a short-term horizon (20 min) to meet the real-time constraints of indoor temperature and hot water supply. To coordinately satisfy the above load demands in two time-scales, in this work we propose a two-stage operation framework for the cruise ship, which is shown as follow:

From the Fig. 7.6, the first stage hourly schedules the DGs, CCHP, and battery to fulfill the voyage distance constraints and hourly electric load demand. The thermal power produced by the CCHP is stored in the thermal energy storage. In the second stage, every 20 min, the PTC and thermal energy storage is dispatched to meet the thermal load demand. With the above operation framework, both the propulsion and electric loads can be met in a long-term time-scale, as well as the thermal load demand can be met in a short-term time-scale to improve the QoS.

Fig. 7.6
figure 6

reprinted from [26], with permission from IEEE

Two-stage operation framework for the cruise ship,

Full size image

To show the benefits of the proposed model, the onboard generation and battery SOC are shown in Fig. 7.7a, b, respectively. Figure 7.8 compares the results of the thermal load of the proposed two-stage method.

Fig. 7.7
figure 7

Onboard generation and battery SOC with/without speed variations, reprinted from [26], with permission from IEEE

Full size image
Fig. 7.8
figure 8

reprinted from [26], with permission from IEEE

Onboard thermal storage and thermal load,

Full size image

From Fig. 7.7a, the battery can coordinate with the speed adjustment to smooth the load profiles, which facilitates the economy of cruise ships (the DGs can better operate around their economic points). From Fig. 7.7b, the battery may have much deeper charging/discharging events without the speed adjustment. That is mainly because the cruising speed is fixed during the cruising time-intervals, and the battery should quickly respond to the load profiles for the economy of navigation.

From Fig. 7.8a, the proposed two-stage scheduling model can meet the thermal load demand in a more accurate time-scale by simply dispatching the loading factor of the thermal storage tank, and the outputs of PTC and CCHP. The results are shown in Fig. 7.8b. The indoor temperature can be kept as a constant meanwhile the single first stage will have a maximal 3 ℃ temperature variations since the accumulated effects of thermal load demand variations. Similarly, the single first stage also cannot meet the hot water supply-demand all the time, and the thermal variations will also be accumulated and make the supplies always smaller than the demands.

Current cruise ships are mainly BOS cruise ships, which means in the BOS mode, the thermal load demand is all provided by the electric-side (PTC units). In this case, the BOS ship replaces the CCHP to conventional DG with the same capacity. The parameters are the same with DG2, 3. The total load demand and EEOI of BOS and HES ships are shown in Fig. 7.9.

Fig. 7.9
figure 9

reprinted from [26], with permission from IEEE

Comparisons between BOS and HES cruise ships,

Full size image

From Fig. 7.9, the BOS cruise ship will have much larger load demands since the thermal load is provided by the PTC unit. Correspondingly, the EEOI of the HES integrated cruise ship is also much smaller than the BOS by 8.37%.

7.4.2 Multi-energy Management for Seaport Microgrids

(1) System description

From Fig. 7.10, there are three energy resources in this microgrid, i.e., photovoltaics(PVs), electrical substation, and gas pressure house. The PVs and substation inject electricity into the seaport microgrid via DC and AC buses, respectively. The gas pressure house injects gas into the seaport microgrid to the gas storage. Additionally, to improve the system flexibility, a battery energy storage system (ESS) and two thermal storages are incorporated. The AC/DC loads and heat/cooling power are supplied to the seaport loads, and DC power is used for charging the electric trunk. The power to gas equipment transforms the excess power to gas to fill the gas vehicles.

Fig. 7.10
figure 10

An illustrated seaport microgrid case revised from [28]

Full size image

In this paper, the scheduling horizon is divided into equal time step \( \Delta t \), denoted by set \( {\mathcal{T}} = \left\{ {1,2, \ldots ,T} \right\} \). The proposed operation method is formulated as a two-stage framework, where the first stage is for the day-ahead time-scale, and the second stage is for real-time scheduling, i.e., hourly. In the day-ahead operation (first stage), the hourly energy scheme is provided considering the uncertainties, and then in the second stage, the seaport microgrid adjusts its scheduling plan responding to the realization of uncertainties in the hourly time-scale. The electrical load profile, heating load profile, and cooling load profile are shown in Fig. 7.11, which are all given in 1000 scenarios. Other detailed parameters can be found in [28].

Fig. 7.11
figure 11

Input parameters of the proposed method, reprinted from [28], open access

Full size image

(2) Case study

To verify the effectiveness of the proposed method, different cases are formulated as follows.

Case 1: Two-stage optimization is considered, meanwhile the joint constraints are considered.

Case 2: Only the first-stage optimization is considered.

(2.1) Bi-directional AC/DC power flow

To show the coordination between AC and DC sides, the power flow via the bi-directional AC/DC converter is shown in Fig. 7.12. The AC to DC power is shown as the surface above the zero surface, while the DC to AC power is shown as the surface below the zero surface. Then, to show the effects of ESS, the state of charge (SOC) of battery is shown in Fig. 7.13.

Fig. 7.12
figure 12

reprinted from [28], open access

Power flow via bi-directional AC/DC converter,

Full size image
Fig. 7.13
figure 13

reprinted from [28], open access

SOC of battery,

Full size image

From the above figure, at first, when the PV power is almost zero, i.e., t = 0–5 h, 20–24 h, the DC load is mainly met by AC to DC converter. When the DC load gradually increases, the AC to DC power is also increasing, and the battery discharges to further support DC load, i.e., t = 5, 6 h. After that, with the PV power increasing, the power demands also become larger, i.e., both DC and AC loads during t = 10–16 h. In those time intervals, the PV output is beyond the maximal DC load, which leads the PV power change to AC via AC/DC converter to support the AC load or charge into battery, which is shown as the surface below zero in Fig. 7.12 and the charging event in Fig. 7.13. From the above results, the integration of the AC/DC converter can bring great flexibilities to meet both DC and AC loads. The DC power for PV and AC power from UG and CHP can coordinately operate to enhance energy efficiency.

(2.2) Multiple energy flows

In this seaport microgrid, various energy carriers are working coordinately to enhance operation flexibility. To show those coordinations, the power of CHP is shown in Fig. 7.14, the power of heat storage is shown in Fig. 7.15, the power of cooling storage is shown in Fig. 7.16, and the power of power-to-gas facility is shown in Fig. 7.17.

Fig. 7.14
figure 14

reprinted from [28], open access

Power of CHP,

Full size image
Fig. 7.15
figure 15

reprinted from [28], open access

Power of heat storage,

Full size image
Fig. 7.16
figure 16

reprinted from [28], open access

Power of cooling storage,

Full size image
Fig. 7.17
figure 17

reprinted from [28], open access

Power of P2G equipment,

Full size image

From Fig. 7.11d and e, there are two demand impulses of both heat and cooling demands in t = 6, 7 h. The CHP responds to those demand impulses and consumes the gas to produce electricity and heat. The heat energy is stored and both the heat and cooling storages are discharging in this period to satisfy the demand, which is shown as the great valleys in their energy curves in Figs. 7.15 and 7.16. After that, CHP is shut-down since the total electricity demand is limited. The thermal demands are then met by the coordination of thermal storage and the gas boiler.

It should be noted that when t = 10–15 h, the temperature increases and requires great air-conditioning power demand. While in this time period, the PV power is also in its peak-hours. Then the PV power is converted to gas for the gas boiler to meet the air-conditioning power demand, which is shown as in Fig. 7.17.

The above results show that different energy carriers can be coordinated flexibly in a seaport microgrid. The excess electricity can be converted to gas for thermal demand. With the interactions between different energy carriers, the electric and thermal demand can both be satisfied and the flexibility can be enhanced.

(2.3) Electric and gas trucks

The energy demand of trucks is quite important in future seaport since they play a major role for cargo lifting and transporting. However, before the completed electrification of vehicles, the gas trunks and electric trunks will both exist in seaport microgrid. To satisfy their energy demands, the electric and gas sub-systems of seaport microgrid should be operated in coordination, respectively. In this case, the equivalent energy of gas trucks are shown in Fig. 7.18, and the charging power of electric trucks are shown in Fig. 7.19.

Fig. 7.18
figure 18

reprinted from [28], open access

Equivalent energy of gas vehicles,

Full size image
Fig. 7.19
figure 19

reprinted from [28], open access

Charging power of electric trunks,

Full size image

From Fig. 7.18, the energy peaks of gas vehicles are t = 10–15 h and 20–24 h. The first peak period corresponds to the working hours, and the second is the vehicles coming back for charging. From the results in Fig. 7.19, the charging patterns are more periodic with three peak hours, i.e., t = 10–15, 16–18, and 20–24 h. From the above results, both the gas and electricity demands of trunks can be satisfied.