https://orcid.org/0009-0001-6912-2108
Abstract
Wind, as well as photovoltaic (PV), is widely used. Like loads, its power cannot be predicted, which results in the grid having to bear the power imbalance between wind-PV and loads, and substantial power fluctuations are not tolerated. Hybrid energy storage systems (HESS) containing multiple storage methods are considered effective solutions. In this paper, pumped storage and lithium-ion battery storage are fully considered, as they are supposed to have excellent performance and are highly complementary. We categorize the power imbalance into low, medium, and high according to the magnitude of the power imbalance. When the power fluctuation is low, the battery dominates. In contrast, the pumped storage dominates when the power fluctuation is high. Most importantly, when the power fluctuation is medium, we utilize an optimized first-order low-pass filter to allocate the power between the pumped storage and the lithium-ion battery. We change the filtering time in real-time according to the battery’s state of charge (SOC) to reasonably allocate the power between the pumped storage and the lithium-ion battery and ensure the SOC fluctuates within a reasonable range. This paper confirms the feasibility of the proposed strategy, where the pumped storage power fluctuates very little, in contrast, the battery power fluctuates significantly, and the SOC is always within the set reasonable range. Most importantly, the strategy proposed in this paper is straightforward to implement, which is crucial for engineering applications.
1. INTRODUCTION
Wind and photovoltaic (PV) are classical renewable energies, often combined due to their strong complementarity [1]. As with loads, their power is unpredictable. The energy storage technology can provide or absorb additional power to realize the power balance of the system when there is a power imbalance between wind-PV and the load, which plays an essential role in the system’s stable operation [2].
There are many different types of energy storage, such as pumped storage, lithium-ion battery, and supercapacitor (SC), each of which has unique characteristics [3]. Currently, the focus is on hybrid energy storage technology because it rationally utilizes multiple energy storage methods to make the system perform better [4]; for example, a hybrid energy storage system (HESS) composed of lithium-ion battery and SC is often focused on because SC has ms-level response time, which can respond to power changes very quickly; however, it stores less energy [5]; the lithium-ion battery does not have the same fast response time as SCs, but can store more energy [6].
Although not as fast as the SC response time, lithium-ion batteries can provide a ms-level to s-level response time, which is sufficient on many occasions, while pumped storage energy storage, with a long life, low cost, mature technology, large capacity, and other advantages, is widely used [7]. However, pumped storage can only provide response time from s-level to min-level, which needs to be improved compared to lithium-ion batteries. It is a good idea to consider combining lithium-ion batteries and pumped storage to form HESS [8].
In the HESS, it is critical to distribute the power of different energy units, and a reasonable way of power distribution can enable the HESS to realize satisfactory performance [9–11]. The commonly used power allocation methods include filter decomposition, wavelet decomposition, empirical modal decomposition (EMD), and variational modal decomposition (VMD) [12, 13]. In [14], the wavelet packet is used to decompose the original power of wind power. The power signal is allocated according to its level, based on which the state of charge (SOC) of HESS is further considered, and the fuzzy logic control is applied to smooth the output of wind power. In [15], model predictive control and first-order low-pass filtering are applied for power allocation. However, the traditional low-pass filtering algorithm has a delay. In [16], the EMD algorithm is used to decompose the raw wind power into low-, medium-, and high-frequency power, and the filtering order is adjusted according to the SOC of the battery and the SC. However, the EMD algorithm has problems such as modal overlapping, endpoint effect, and difficulty determining the stopping condition.
In this paper, we propose a simple and easy-to-implement control strategy to rationally allocate power based on pumped storage and a HESS composed of lithium-ion batteries, and we would like to obtain a strategy that is easier to implement because more straightforward methods have higher reliability and stability.
2. CONTROL STRATEGY
Due to the unpredictability of the wind-PV power and the load power, when their power is not matched, it leads to fluctuations in the bus power, and the HESS needs to provide energy to the load or absorb the wind-PV power to achieve power matching.
Pumped storage usually has a considerable capacity, costs much less than lithium-ion batteries, and has a long service life, often smoothing out significant power fluctuation. Lithium-ion batteries are characterized by a much faster response time than pumped storage, but their small capacity can only smooth out small power fluctuations. This paper is based on the characteristics of lithium-ion batteries and pumped storage and the control strategy design shown in Figure 1.
Control strategies for HESS.
As shown in Figure 1, the power fluctuation between the load and the wind-PV is categorized into three levels, i.e., small, medium, and high, and these three different levels of power fluctuation will be used with three different control strategies. Although many techniques, such as fuzzy logic control, have been proposed, it is further subdivided to improve the performance; in fact, the solution we use has sufficient performance and is very simple and well-suited for industrial applications.
2.1. Strategy A
Figure 2 shows the control strategy for strategy A. Since it is only a slight power fluctuation, it is usually unnecessary to activate pumped storage, and it is sufficient to rely solely on the batteries. Lithium-ion batteries dominate, and pumped storage only plays a supporting role. However, when the SOC of the battery is low, if the wind-PV power is less than the load power, and the HESS needs to provide more power to the load, then pumped storage must be activated to charge the SOC of the battery up to 50%, and then stop, during this process, even if the bus power changes, for example, the wind-PV power is greater than the load power, the pumped storage should not be stopped until the battery SOC reaches 50%. Similarly, when the battery SOC is high, the wind-PV power is greater than the load power. The HESS must absorb additional energy, and pumped storage must also be activated to maintain the battery SOC at 50%. The battery needs to release extra power to ensure that it can absorb the wind-PV power.
Control strategies for strategy A.
2.2. Strategy B
Strategy B is the most complex because, at this time, the need for lithium-ion batteries and pumped storage work simultaneously. How to reasonably allocate the power of lithium-ion batteries and pumped storage so that they can fully play their respective advantages is the key to the problem. Usually, the unbalanced power between wind-PV and load can be decomposed into high-frequency and low-frequency power. Lithium-ion batteries have a breakneck response speed, which is highly suitable for responding to high-frequency power, so it’s a good idea to allocate high-frequency power to lithium-ion batteries and low-frequency power to pumped storage.
2.2.1. Common strategy
As shown in Eq. (5), in response to the high-frequency power command, the first-order low-pass filter has an integral effect independent of the PHESS(t), which makes the battery have a capacity accumulation effect. This will very easily lead to over-charging or over-discharging of the battery, which affects the operation of the HESS.
2.2.2. Optimization strategy
Based on the standard first-order low-pass filter analysis, we optimize its filtering time T according to the SOC of the lithium-ion battery to avoid over-charging or over-discharging the battery pack.
As shown in Figure 3, the SOC of the battery is considered medium when it is in the range SOCminL − SOCmaxH, above which it is considered high, and below which it is considered low. When the SOC of the battery is medium, it is divided into two additional regions, SOCminL − SOCminH and SOCmaxL − SOCmaxH. When the SOC of the battery is in these two regions, it means that its SOC will deviate from the normal range, and then it is necessary to change the filtering time to T1 to make the SOC of the battery return to the normal range.
Schematic diagram of zone division according to SOC.
Please note that the following analyses are all based on a charged battery. We develop the following control strategies according to the battery SOC:
(1) 0 − SOCminL
The battery SOC is low, and the battery pack is absorbing extra power from the wind-PV, a more significant filtering time can be taken into account, allowing the battery to absorb more energy to increase its SOC to a reasonable range, assuming that the maximum filtering time is Tmax, then T1 = Tmax.
(2) SOCminL − SOCminH
During the charging process, as the SOC of the battery rises, the filtering time T1 turns from the maximum value Tmax to T, which accordingly causes part of the fluctuating power initially absorbed by the pumped storage to be transferred to be absorbed by the lithium-ion battery.
(3) SOCminH − SOCmaxL
The battery SOC is optimal in this interval and does not require additional work, so it is only necessary to maintain its filtering time, then T1 = T.
(4) SOCmaxL − SOCmaxH
In this case, as the SOC of the battery rises, part of the fluctuating power borne by the original battery is continuously transferred to the pumped storage, and the pumped storage needs to absorb the energy that the lithium-ion battery should have absorbed, so the filtering time, T1, decreases with the increase of the SOC of the battery. When the SOC of the battery reaches the upper limit, the filtering time is correspondingly reduced to 0, and the pumped storage bears the fluctuating power.
(5) SOCmaxH − 100%
In this case, the battery SOC is high and cannot continue to absorb the extra power from wind-PV; all the energy should be absorbed by pumped storage, then T1 = 0.
2.2.3. Performance analysis
As shown in the above analysis, we have divided the control strategy into five types according to the SOC of the lithium-ion battery; when the battery is charging, as its SOC rises, more and more power is transferred to the pumped storage, until finally all the energy is absorbed by the pumped storage, which better avoids the danger of over-charging, for the discharge the principle is the same.
While SOCmaxH = 80% and SOCmaxL = 70%, the variation between a and b is shown in Figure 4; in this paper, we chose a = 2 because compared to a = 1, it performs better. When the battery SOC rises from 70% to 80%, the a = 2 has a slower response in the initial stage; that is, when the battery is in the SOC warning area, a = 2 still allocates more power to the battery, but as the battery SOC rises again, the power assigned to the battery will decrease dramatically, which has the advantage that it can fully utilize the SOCmaxL − SOCmaxH because battery SOC is ubiquitous in this interval and is not intolerable. According to the principle of the first-order low-pass filter, the battery should absorb the power, so it is unnecessary to take significant action when the battery SOC is just over 70%. Furthermore, overall, the battery will absorb more power at a = 2 than at a = 1, which fully utilizes the capacity of the battery, and ultimately, the battery will not be allowed to charge when SOC >80%.
The variation between a and b in Eq. (7).
a = 3 or a = 4 have a similar effect to a = 2; however, their b is almost 0 when SOC <74%, meaning that the power of the battery is gradually reduced almost until SOC >74%, and they are reduced too much, with more significant battery power fluctuations compared to a = 2, which affects the performance of the HESS, so a = 2 is the excellent solution.
Strategy B is easy to implement; its control flow can be easily programmed, and there is little possibility of malfunction, while the excellent performance of the first-order low-pass filter is well preserved, which is highly beneficial for industrial applications.
2.3. Strategy C
Figure 5 shows the control strategy for strategy C. With large power fluctuations, it is sensible to utilize pumped storage to smooth out the power difference between the wind-PV and the load since low-capacity lithium-ion batteries cannot carry large power fluctuations. However, until then, the battery will be activated because it has a speedy response time and will be decommissioned when the pumped storage power reaches the required level. However, when the SOC of the battery is too low, the pumped storage will adjust its power to maintain the SOC of the battery up to 50%; for example, when the wind-PV power exceeds the load power, the battery will absorb some of the overflow power and the pumped storage will absorb the rest of the power to increase the battery SOC. Similarly, when the SOC of the battery is too high, the pumped storage will regulate its power, allowing the battery to release some of its energy to ensure that it can easily cope with bus power fluctuations.
Control strategies for strategy C.
3. CASE STUDY
In this paper, we assume that the capacity of pumped storage in HESS is 100 MWh, and the capacity of lithium-ion batteries is 250 kWh. For the assignment of battery SOC ratings, the battery SOC is considered high when the SOC is higher than 80% and low when the SOC is lower than 20%. In the case of a medium SOC, the filtering time does not need to be altered in the range of 30%–70%, while in other intervals, it needs to be adjusted in real-time according to its SOC. The power fluctuation is low when the power fluctuation between the wind-PV and the load is lower than 250 kW. When the power fluctuation is higher than 750 kW, the power fluctuation is high. The rest is considered medium power fluctuation, which the lithium-ion battery and the pumped storage must smooth out. For ease of analysis, positive values represent HESS absorbed power, and negative values represent HESS released power.
3.1. Case 1
Case 1 is concerned with the situation where lithium-ion batteries or pumped storage alone provide the power, which is more straightforward because no complex control is required. Here, we only discuss the small power fluctuation case because the analysis of the enormous power fluctuation is the same as the slight power fluctuation. Figure 6 shows the power imbalance between wind-PV–load and the output power of pumped storage and battery, Figure 7 shows the battery SOC variation. According to the logic of strategy A, only the lithium-ion battery is needed to smooth out the power fluctuation, and the pumped storage does not need to work. However, at ~80 min, the pumped storage starts and absorbs power, and the source of this power includes the battery; the battery is supplying energy to the pumped storage, which is because the battery SOC has exceeded 80% and reached its limit, and the pumped storage always works until the battery SOC is 50%, although the power of the wind-PV–load is constantly changing during the process, including the HESS supplying power as well as absorbing energy. At ~300 min, the battery SOC is reduced to 50%, then the battery can cope with charging and discharging with better performance, so the pumped storage stops working, and the subsequent work is basically according to the logic of strategy A.
HESS power fluctuations.
Battery SOC variation (case 1).
3.2. Case 2
Figure 8 shows the power imbalance between the wind-PV and the load.
Power imbalance fluctuation between wind-PV and loads.
Figure 9 to Figure 11 show the pumped storage power fluctuations, lithium-ion battery power fluctuations, and battery SOC variation with optimization and without optimization, respectively. Before ~500 min, the power fluctuation of pumped storage and lithium-ion batteries is almost the same because, at this time, the battery SOC is in the range of 30%–70%, and the control logic of both strategies is the same. However, as the battery SOC continues to decline until it is below 30%, the strategy without optimization does not make corresponding actions. The power fluctuations of its pumped storage and lithium-ion batteries are not different from those of the previous one, which results in the battery SOC continually decreasing at a minimum of ~15%, which is then very easy to make the battery over-discharge and affect the safe operation of the system. However, the strategy is optimized; as the battery SOC decreases, the pumped storage correspondingly reduces the absorbed energy limit, allowing the battery to absorb more power to maintain its SOC. Eventually, as shown in Figure 11, the battery SOC is at a minimum of ~26%, which achieves the goal well.
Pumped storage power fluctuations.
As shown in Figure 9 and Figure 10, for the strategy with optimization, the power fluctuation of pumped storage is smaller than that of the battery. However, the power of pumped storage fluctuates considerably at ~500min, ~700min, and ~1000min to maintain the battery SOC in the normal range, and this fluctuation is cost-effective in getting the battery back to safe operation. In the other cases, the power fluctuation of pumped storage is tiny. In comparison, the power fluctuation of lithium-ion batteries is much more significant; with the faster response time of lithium-ion batteries, this performance can give full play to its advantages, HESS also hopes to achieve this goal.
Lithium-ion battery power fluctuations.
Battery SOC variation (case 2).
As shown in the analysis of case 1 and case 2, the strategy proposed in this paper is performance with expectations, better utilizes the characteristics of lithium-ion batteries and pumped storage, smooths out the power fluctuation between the wind-PV and the load, and ensures the safety of the battery as permissible, in addition to the fact that it is straightforward to implement and has a high degree of reliability.
4. CONCLUSION
It is feasible to smooth the power fluctuation between wind-PV and load using pumped storage and HESS consisting of lithium-ion batteries. By categorizing the power fluctuations into low, medium, and high, it is practical to use different strategies to smooth the power fluctuations; depending on the gap between the battery SOC and the set SOC, it is feasible to avoid over-charging or over-discharging the battery by reversing it to affect the filtering time of the first-order low-pass filter.
Credit Author statement
Lile Wu (Project administration [Equal], Software [Equal], Writing—original draft [Equal]), Huanran Wang (Formal analysis [Equal], Methodology [Equal], Writing—review & editing [Equal]), Zutian Cheng (Data curation [Equal], Software [Equal]), Lei Bai (Writing—original draft [Equal], Writing—review & editing [Equal]), and Helei Li (Writing—review & editing [Equal])
DATA AVAILABILITY
Data is available upon reasonable request.
