Optimal Residential Load Control With Price Prediction in Real-Time Electricity Pricing Environments
Summary (6 min read)
Introduction
- Wholesale electricity market, real-time pricing, inclining block rates, price prediction, energy consumption scheduling.
- Time-differentiated pricing is currently implemented in various regions in North America, e.g., in form of hourly-based DAP tariff used by the Illinois Power Company in the U.S. [14] and the three-level (on-peak, mid-peak, off-peak) TOUP tariff used by the Ontario Hydro Company in Toronto, Canada [15].
- Recent studies have shown that despite several advantages that RTP and IBR can offer, the lack of knowledge among the users about how to respond to time-varying prices and the lack of effective home automation systems are two major barriers for fully utilizing the benefits of real-time pricing tariffs [20], [21].
- The rest of this paper is organized as follows.
II. SYSTEM MODEL
- The authors provide a mathematical representation of the residential load control problem in RTP environments with IBR.
- The authors consider the general wholesale electricity market scenario shown in Fig. 1, where each retailer/utility serves a number of end users.
- The RTP information, reflecting the wholesale prices, are informed by the retailer to the users over a digital communication infrastructure, e.g., a local area network (LAN).
- Their focus is to formulate the energy consumption scheduling problem in each household as an optimization problem that aims to achieve a trade-off between minimizing the electricity payment and minimizing the waiting time for the operation of each household appliance in response to the real-time prices announced by the retailer company.
- The authors will explain how the optimization problem in this section can be solved in practice later in Section IV.
A. Residential Consumers
- Consider a residential unit that participates in a real-time pricing program.
- In case of a plug-in hybrid electric sedan, in total Ea = 16 kWh is needed to charge the battery for a 40-miles driving range [2].
- Next, assume that for each appliance a ∈ A, the user indicates αa, βa ∈ H as the beginning and end of a time interval in which the energy consumption for appliance a is valid to be scheduled, respectively.
- On the other hand, if βa − αa ≫ 3, it is possible to select certain hours within the large interval [αa, βa] to schedule energy consumption such that the electricity payments can be minimized.
- More details about various home area network technologies can be found in [32].
B. Real-Time Pricing with Inclining Block Rates
- Recall from Section I that RTP and IBR are two promising non-flat pricing models to replace the current flat rate tariffs.
- The authors provide a general mathematical pricing representation which combines these two pricing models.
- This is indeed the case in DAP structures.
- Let lh , ∑ a∈A x h a denote the total hourly household energy consumption at each upcoming hour h ∈ H. Recall that H denotes the scheduling horizon.
C. Problem Formulation
- Given the feasible energy scheduling set X and the RTP model in (6), the key question is:.
- As an alternative, the user might be willing to wait for 2 hours only and save 1.5 cents instead.
- Next, the authors explain how this trade-off can be mathematically taken into account in an optimization-based framework.
- From the RTP model introduced in Section II-B, the user’s total electricity payment corresponding to all appliances within the upcoming scheduling horizon is obtained as H∑ h=1 ph (∑ a∈A xha ) × (∑ a∈A xha ) , (11) where the price function ph(·) is as in (6).
- On the other hand, parameter δa acts as a knob to control the trade-off between the two design objectives with respect to minimizing the payment and the waiting cost for each appliance.
III. PRICE PREDICTION IN REAL-TIME PRICING ELECTRICITY ENVIRONMENTS
- So far, the authors have assumed that each end user is fully aware of the upcoming price values set by the utility company within the scheduling horizon H .
- That is, the user always knows the values of ah, bh, and ch for each h ∈ H.
- The authors may consider more dynamic pricing scenarios where the upcoming prices are announced only for 1 ≤ P ≪ H hours ahead of time.
- Clearly, the extreme case would be P = 1 when only the next hour price is released.
- In these cases, any energy consumption scheduling policy, including the optimization-based energy consumption scheduling approach described in Section II-C, essentially requires some price prediction capabilities.
A. Prediction Based on Prior Knowledge
- In general, price parameters may depend on several factors.
- From the results in this figure, the authors can see that although the trends are partially (not exactly) similar in different years, the exact prices can be drastically different.
- This suggests that there can be relationships between the upcoming prices and whether the prices are for a working day or for a weekend.
- Let âh[t], b̂h[t], and ĉh[t] denote the predicted parameters for the upcoming price tariff for each hour h on day t.
- The optimal choices of the prediction filter coefficients are shown in Table I.
IV. OPTIMAL RESIDENTIAL LOAD CONTROL
- Recall from Section II-C that problem (15) is not tractable in its current form due to the non-differentiability of price function ph(lh) in (6).
- The authors explain how one can solve problem (15) in practice.
- The first part shows the exact payment within the next P hours while the second part shows the estimated payment within the H−P hours after that.
- Clearly, same as that in problem (15), the objective function in optimization problem (19) is non-differentiable.
V. REMARKS, SPECIAL CASES, AND EXTENSIONS
- The proposed optimization-based residential load control framework in this paper can be extended in various directions.
- The authors overview a number of scenarios that can be addressed by slight modification in the system model.
A. Appliances with Discrete Energy Consumption Levels
- Recall from Section II-A that in their system model, the hourly-based energy consumption scheduled for each appliance is a continuous variable which is lower-bounded by γmina and is upper-bounded by γ max a .
- The modified version of problem (25) would be a linear mixed integer program which is more complicated than a linear program, but still can be solved by using optimization software such as CPLEX [35] and MOSEK [36].
- B. Interruptible and Uninterruptible Residential Load Some load such as charging the battery for a PHEV are interruptible.
- That is, it is possible to charge the battery for one hour, then stop charging for another hour, and then finish the charging after that.
- Again, considering the case with discrete energy consumption levels as in Section, V-A, for each uninterruptible appliance a, let θa denote the duration of time, in number of hours, that appliance a needs to operate at power level γmaxa .
C. Availability of Multiple Retail Electricity Sources
- Consider the case where an end user has the ability to obtain electricity from a set of S utility companies simultaneously, where |S| > 1.
- Similarly, the authors define x h a,s as the energy consumption scheduling variable corresponding to each appliance a for energy consumption obtained from each utility s at each time h.
- That is, it would be optimal for the end user to obtain all its energy need from the utility with lowest price.
- If IBR is adopted, then obtaining the optimal solution would be more complicated as it could be beneficial for the user to distribute its load among the available utility companies to avoid being charged with the rates at the higher block.
- Nevertheless, the reformulated version of optimization problem (25) for the case with multiple utility companies would still be a linear program.
D. Avoiding Load Synchronization
- In general, it is desired for the power distribution and wiring systems such that no load synchronization occurs among different appliances in each household.
- As the authors will see in Section VI-D, adopting IBR model helps in avoiding the concentration of a large portion of energy consumption in a single low-price hour.
- On the other hand, it is also desired to even avoid synchronization among different appliances that start operation exactly at the same hour in order to prevent a sharp spike in the residential load.
- This can be done by introducing a short random starting delay, e.g., a few seconds, to diversify the starting moments among different appliances.
E. Announcing the Scheduled Consumption Back to the Utility
- One of the main challenges that the utility companies face is the need for predicting the demand load by end users.
- Clearly, by knowing the upcoming demand, the utility companies and regional power plants can better perform energy dispatching.
- Such predictions may only be done statistically in the current electric grid.
- By large deployments of the automatic residential load control strategies that the authors proposed in this paper, the end users are potentially capable of announcing their upcoming load back to the utility company through a twoway digital communication infrastructure as in the one already shown in Fig.
- Given the expected load from all users, the utility company would have an accurate estimation of the load that it needs to provide within the next couple of hours.
F. Handling Load Reduction Requests
- Load reduction requests are usually sent out by the utilities when electricity demand is high enough to put the grid reliability at risk, or rising demand requires the imminent activation of expensive/unreliable generation sets.
- In a smart grid, an advance notice for load reduction can be sent through the communication infrastructure to each meter asking the energy scheduler to take an appropriate action.
- This can be done in their design by increasing the prices used in the optimization for the next two or three hours.
- This will automatically postpone a portion of the upcoming energy consumption to some later hours leading to a major decrease in the total load.
G. Residential Electricity Storage
- As PHEVs become popular, there is an increasing interest in using the storage capacity of their batteries to return some energy back to the grid when needed [37].
- The users would buy electricity for charging their batteries at some low-price hours and then sell electricity back to the grid by discharging their batteries when the price is high.
- Therefore, the users can not only help balancing supply and demand in the regional electricity market, but also make money.
- Due to the same reasons discussed in Section I, it is difficult for the users to keep monitoring the real-time prices in order to decide when it is the best time to charge or discharge their batteries.
- The proposed optimization-based load control model can be extended in this regard by incorporating negative loads for discharging actions.
H. Accommodating Changes in Users’ Energy Needs
- The energy scheduler discussed in this paper can update schedules at any time based on the user’s needs.
- Assume that an initial energy schedule is planned at 8:00 AM and the appliances are assigned to consume energy accordingly.
- In that case, the energy scheduler can adjust the existing choices of energy consumption schedules and solve problem (25) based on the new situation.
- Clearly, for those appliances that are already in the middle of their operation, parameter Ea is recalculated accordingly.
VI. SIMULATION RESULTS
- The authors present the simulation results and assess the performance of their proposed residential load control scheme with price prediction.
- Unless the authors state otherwise, the simulation setting is as follows.
- For the purpose of their study, the authors assume that the number of appliances used in this household at each day varies from 10 to 25.
- The authors assume that the scheduling horizon H = 24.
- On the other hand, although the pricing model used by IPC is day-ahead (i.e., price announcement horizon P = 24), the authors also address the cases where P ≪ 24 which are better representations of the “real-time” pricing tariffs.
A. Performance Gains from Users and Utility Prospectives
- The authors start by looking at the resulting payments from the users as well as the PAR in the residential load when they use their proposed load control model.
- The trends of daily electricity charges and PAR for a sample residential load based on the day-ahead real-time prices adopted by IPC from September 1 to December 31, 2009 are shown in Fig.
- From the results in Fig. 8(a), by using the proposed load control scheme the user’s average daily payment decreases by 25% from 108 cents to 81 cents.
- For the experiment studied in Section VI-A, the authors assumed that the price announcement horizon P = H = 24 hours.
- That is, the authors considered the day-ahead pricing model.
C. The Impact of Scheduling Control Parameter δa
- Recall from Section II-C that the user can balance payment and waiting time for the operation of each household appliance by adjusting parameter δa for each appliance a.
- As δ increases it will also be desired to finish the operation of each scheduled appliance sooner.
- The results on monthly electricity payment and average waiting time are shown in Fig. 10.
- As the authors increase δ, the charges will increase while the waiting times decrease.
- Of course, it is entirely up to the user to decide if he prefers paying less or instead getting the work done by the appliances within a shorter period of time.
D. The Impact of Adopting Inclining Block Rates
- The authors show that combining RTP with IBR is indeed helpful in achieving more balanced residential load and avoiding load synchronization.
- The authors can see that an RTP tariff would lead to high PAR due to congestion at low-price hours.
- In fact, without IBR, the optimal solution of problem (25) is nothing but scheduling most of all energy consumption at hours with lower prices.
- This problem may not be visible in the existing manual residential load control programs as it is not easy or even possible for the users to keep watching the prices and only turn on their appliances when the prices are low [20], [21].
- In an automatic load control such a congestion scenario can occur frequently and become troublesome.
E. The Impact of Number of Users
- So far, the authors have focused their simulation studies on scenarios with only a single user.
- The authors consider the case when a utility company serves 10 users which are all equipped with their proposed automatic residential load control capability.
- The simulation results are shown in Fig. 12.
- The authors can see from the PAR values in Fig. 12(a) that increasing the number of users can further balance the aggregated load even if no load control strategy is being deployed.
- On the other hand, all users still pay less on their monthly electricity bills.
VII. CONCLUSIONS
- Its benefits are currently limited due to lack of efficient building automation systems as well as users’ difficulty in manually responding to time-varying prices.
- The authors focused on a scenario when real-time pricing is combined with inclining block rates in order to have more balanced residential load with a low peak-to-average ratio.
- By applying a simple and efficient weighted average price prediction filter to the actual hourly-based prices adopted by Illinois Power Company from January 2007 to December 2009, the authors obtain the optimal choices of the coefficients for each day of the week.
- Simulation results show that the combination of the proposed energy scheduler design and the price predictor leads to significant reduction in users’ payments.
- This encourages the users to participate in the proposed residential load control program.
Did you find this useful? Give us your feedback
Citations
2,715 citations
Cites background from "Optimal Residential Load Control Wi..."
...While it is usually difficult and confusing for the users to manually respond to prices that are changing every hour [16], [17], another problem that RTP may face is load synchronization, where a large portion of load is shifted from a typical peak hour to a typical nonpeak hour, without significantly reducing the PAR [18]....
[...]
2,433 citations
2,337 citations
Cites background or methods from "Optimal Residential Load Control Wi..."
...Mohsenian-Rad and Leon-Garcia [170] proposed an optimal and automatic residential energy consumption scheduling framework, which attempts to achieve a desired trade-off between minimizing the electricity payment and minimizing the waiting time for the operation of each appliance....
[...]
...For optimization approaches, the commonly used mathematical tools are convex programming [120], [170], [220], [228] and dynamic programming [8], [95], [130], [172]....
[...]
...Individual User [32], [40] [53], [69] [95], [108] [120], [171] [170], [188] Multiple Users [98], [220] Electricity Industry [37], [74] [83], [89] [120], [150] [171], [178] [206], [218] Price [213] Emission [14], [32] [86], [151] [218]...
[...]
...[171] A.-H. Mohsenian-Rad, V. W. S. Wong, J. Jatskevich, R. Schober, and A. Leon-Garcia....
[...]
...[170] A.-H. Mohsenian-Rad and A. Leon-Garcia....
[...]
1,901 citations
1,014 citations
Cites background from "Optimal Residential Load Control Wi..."
...There exists a large literature on demand response, see, e.g., [1], [2], [3], [4], [5], [6], [7], [8], [9], [10]....
[...]
...Besides the work such as [4], [5] which considers a single household demand response given a pricing scheme, [6] considers a power network where end customers choose their daily schedules of their household appliances/loads by playing games among themselves and the utility company tries to adopt…...
[...]
References
[...]
33,341 citations
2,780 citations
"Optimal Residential Load Control Wi..." refers background in this paper
...Therefore, it can be solved efficiently by using linear programming techniques [33]....
[...]
[...]
1,661 citations
"Optimal Residential Load Control Wi..." refers background in this paper
...In particular, during the charging time, the PHEVs double the average household load [2]....
[...]
...needed to charge the battery for a 40-mi driving range [2]....
[...]
...For example, a PHEV may be charged only up to per hour [2]....
[...]
...For example, for a single-phase PHEV, the normal charging time is 3 h [2]....
[...]
...3 kWh of charging power for one mile of driving [2]....
[...]
516 citations
"Optimal Residential Load Control Wi..." refers background in this paper
...Clearly, the user is interested in reducing its charges while the utility is interested in having a balanced load demand with a low PAR....
[...]
...However, we can reduce PAR by increasing the price at the higher block in IBR....
[...]
...While the optimization problem we study in this paper is partly similar to the one studied in [17] for fixed prices, here we take into ac-...
[...]
...We can see that an RTP tariff would lead to high PAR due to congestion at low-price hours....
[...]
...Last but not least, price prediction is not studied in [9], [17]....
[...]
445 citations
"Optimal Residential Load Control Wi..." refers background in this paper
...The earliest peak-load pricing discussion dates more than half a century ago [6], [7]....
[...]
Related Papers (5)
Frequently Asked Questions (9)
Q2. What is the common scenario for a load reduction request?
Load reduction requests are usually sent out by the utilities when electricity demand is high enough to put the grid reliability at risk, or rising demand requires the imminent activation of expensive/unreliable generation sets.
Q3. What is the main reason why PHEVs are becoming popular?
As PHEVs become popular, there is an increasing interest in using the storage capacity of their batteries to return someenergy back to the grid when needed [37].
Q4. What are the benefits of real-time pricing?
Although real-time pricing has several potential advantages, its benefits are currently limited due to lack of efficient building automation systems as well as users’ difficulty in manually responding to time-varying prices.
Q5. How can the authors improve price prediction accuracy?
Price prediction accuracy can further improve by using longer and more computationally complicated price prediction filters, if needed.
Q6. How long does it take to charge a PHEV?
As another example, the user may select αa = 10 PM and βa = 7 AM (the next day) for his PHEV after plugging it in at night such that the battery charging finishes by early morning time when he needs to use the vehicle to go to work.
Q7. How can the user balance payment and wait time for each appliance?
Recall from Section II-C that the user can balance payment and waiting time for the operation of each household appliance by adjusting parameter δa for each appliance a.
Q8. Why is the price function ph(lh) not tractable?
Note that optimization problem (15) is not tractable in its current form due to the non-differentiability of the price function ph(lh) in (6).
Q9. What is the reformulated version of optimization problem for the case with multiple utility companies?
the reformulated version of optimization problem (25) for the case with multiple utility companies would still be a linear program.