- Magnus Korpås
- WP5 Lead
- NTNU IEL
Flexible resources in the power system (WP5)
The overlaying objective of WP5 is to develop methods and models for cost effective integration of flexible resources in smart distribution grids.
The expected impact of the research is to contribute to improved efficiency of the system operation when utilising flexibility as an important asset in the power system.
- Grid flexibility categorisation and modelling
- Methods for sizing and placement of flexible resources
- Customers' involvement in flexibility
- Value of flexibility
We develop methods and models for cost effective integration of flexible resources in smart distribution grids. They will improve the efficiency of the system operation, when flexibility is utilized as an alternative to grid investments and serving flexibility to the transmission level.
As "Flexibility" was a common focus area in CINELDI in 2020, we increased our integration with the other WPs.
- The integration of flexible resources in active distribution planning (Co-op with WP1)
- The impact of flexible resources on power system security (Co-op with WP1)
- The role of the end-users in the utilization of flexibility (Co-op with WP3)
- Utilization of local battery in Microgrid operation, linked to the Skagerak pilot (Co-op with WP4)
- Drafting of a White Paper on flexibility, to be completed in 2021 (Co-op with all WPs)
Computer models for shiftable atomic loads
If we are to build a more flexible grid, modelling the flexibility potential in the distribution system is crucial. We need data from the models as input for network planning, flexibility aggregation and distribution network operational planning activities.
In 2020 we developed a model for shiftable atomic (uninterruptable) loads. They are shiftable because you can start them any time during the day, but once they start, they cannot be interrupted. The model can be used to estimate aggregated flexibility potential from a group of households at any time of the day. Network operators can use it in both their operation and long-term plans. Aggregators and market participants can estimate the available flexibility from these appliances in the daily operation of their assets.
The model is data-driven, using statistical data and other previously available time series measurements. The model can be updated whenever time series and statistical data is available.
The models are developed by SINTEF Energy Research in the in-kind project KPN ModFlex. The model is available on Zenodo.
Model for operation planning of flexible resources in buildings
As part of his PhD-project (co-financed by CINELDI and FME ZEN), Kasper Thorvaldsen has developed a model which finds the value of flexibility for longterm operation of buildings. Thorvaldsen won the Roy Billinton Best student paper award, gold, at the PMAPS-conference 2020.
The model captures the future (uncertain) impact of current decision making and is inspired by water value calculation in hydro power. It considers both longer periods (from days to months) and uncertainty.
Including the future impact of current decision-making within buildings, energy system scheduling can be crucial when future long-term operational costs are considered. If the future long-term value is not included in a short-term setting, the operational planning can be inaccurate for the total picture.
The model can be further developed into a practical operational tool for scheduling of building energy systems. The long-term planning can be combined with a short-term operational model so that both the short-term and an overview of the future is also considered. It can also be utilized by grid companies who want to study in detail how flexible end-users can respond to different grid tariffs and grid limitations.
High-Performance Multi-Period AC Optimal Power Flow Solver
AC Optimal Power Flow (AC OPF) is a necessary tool for modern grid planning, as wind, PV, energy storage and flexible demand become common parts of the system.
But energy storage and flexible demand makes AC OPF computational very challenging to solve. Computation time is also an issue when using commercial or free optimization solvers.
We have developed a new model which is able to solve the Multi-Period AC OPF problem in a fast way, making it highly attractive for simulation of large and complex grids. The method has been successfully tested on different test grids with different sizes and complexity.
The tool is relevant for DSOs facing new challenges in planning and operation of their grid, such as
- Increasing amounts of prosumers with PV and batteries. The grid operators must be able to predict their net load profile and give the right price or control signals for activating use of flexibility for grid services.
- Increasing amounts of medium-scaled distributed generation, such as smaller wind farms and solar PV farms. These can be in areas where the grid is weak. Energy storage can be an alternative to grid reinforcements.
We have developed an empirically anchored analysis of the concepts of ‘flexibility capital’. Having flexibility capital entails both owning technologies and using electrical loads that can be flexibly managed.
Affluent energy users are more likely to own energy technologies that afford flexibility (such as batteries and smart appliances) and consequently have significant loads that are possible to manage.
Less affluent energy users are less likely to own such technologies that can act as buffers between their daily practices and the flexibility adjustments. Consequently, their flexibility capital is mostly derived from changes to daily activities and routines.
This work adds to improve the understanding of how different grid customer groups can and will contribute with flexibility, and how DSOs and other stakeholders can utilize this knowledge to improve their services and profitability.
The overall objective of WP5 is to develop methods and models for cost effective integration of flexible resources in smart distribution grids.
We expect this to contribute to improved efficiency of the system operation when utilizing flexibility as an important asset in the power system – as a realistic alternative to grid investments and serving flexibility to the transmission level.
Incorporating energy storage and variable renewables in power flow analysis
The increasing penetration of wind power and solar PV introduces challenges to distribution grids, since these sources are associated with variability and uncertainty on multiple time scales, both within each hour, within each day, and over the year. Distribution system operators (DSOs) may therefore have to consider new flexible resources in both the operation and planning of the distribution system.
Energy storage is a class of flexible resources that has received considerable attention lately, by the research community as well as by system operators and end-users. The introduction of energy storage implies a multi-period operational planning problem since an amount of energy discharged by the energy storage at one time step will have to be charged at a previous time step. When the optimal operation of a distribution system is formulated as an alternating current (AC) optimal power flow (OPF) problem, the time-coupling introduced by grid-connected energy storage therefore transforms the problem to a multi-period OPF (MPOPF) problem over an operational planning horizon comprising multiple time steps.
We have therefore done research in 2019 to make sure that energy storage and variable renewables are taken into account in a realistic way in power flow analysis. Our approach is to explicitly take account for uncertainties in future generation and load to ensure a best possible operation decision for the energy storage, inspired by how hydropower producers plan their usage of stored reservoir water. The methods and models that we have developed are relevant for DSOs who are facing different challenges in planning and operation of their grid, such as:
- Increasing amounts of prosumers connected to their grids, with home PV and batteries. The grid operators must be able to predict their net load profile, and also give right price or control signals for activating use of the home batteries for grid services.
- Increasing amounts of medium-scaled distributed generation, such as smaller wind farms and solar PV farms. These can be located in areas where the grid is weak. Energy storage can be an alternative to grid reinforcements. Our models can be used to assess the storage alternative in a robust and effective way.
- Increasing demand for electricity such as high- power charging of electric vehicles and ferries. In some cases, the vehicles themselves represent a source of flexibility, through smart charging and possible Vehicle-to-Grid (V2G) operation. In other cases, the vehicles need high amounts of power in a short time, and stationary energy storage can then be a viable alternative to costly grid reinforcements.
We have published the model and results from a Norwegian test case in a journal paper in 2019, and there are many possibilities for extensions and uses that are relevant for grid planners, such as modelling uncertainty in load due to charging of electrical vehicles, microgrid applications, grid congestions, outages, and/or sharp load peaks. Accounting for the expected value of lost load in setting the end-value of stored energy is particularly relevant for applications where energy storage is used for improving the reliability in constrained grids.
The presented model is a useful tool for grid owners who needs to investigate the integration of renewable energy sources, storage and flexible demand, taking into account both grid details, economics and the stochatic nature of wind, solar and demand.
Methods for cost benefit analysis of batteries in distribution grids
Battery Energy Storage Systems (BESSs) can be used for a range of applications in the power systems, such as load leveling, balancing of variable renewable energy sources (VRES), offer various ancillary services and transmission & distribution grid deferral or a combination of applications. For the latter application, BESS' can be deployed at strategic locations of the transmission & distribution network and perform active and reactive power control for better utilization of the existing grid, as an alternative to costly structural upgrades. This usage of a BESS is relevant for areas with expected growth in demand, power quality issues and/or integration of a large amount of VRES.
However, since grid applications of BESS' are still in the early stages of deployment, there is still no consensus on recommended computational methods for performing cost-benefit analysis of BESS' as alternative for grid upgrades, or for other grid services. In contrast to traditional static grid assets (e.g. transmission lines), the benefits of BESS' depend upon the dynamics and the control strategy of how these are operated. Especially in grids with large amounts of VRES, the storage dynamics increase the system complexity and thereby will require more advanced computational methods for grid planning than presently employed in practice. The lack of established computational methods for including BESS' in grid planning is a barrier for taking published research-based models into practice.
In 2019 we therefore performed a comprehensive and systematic overview of relevant computational methods reported in the literature.
We found it will be crucial for the grid operators to be able to include BESS in their planning procedures in a proper way, whether they are allowed to install BESS themselves or will rely on other actors to provide grid services. We find that for the research-based methods to be suitable for grid planning, several issues need to be accounted for: The methods should handle timing of installations as well as sizing and siting, especially for cases where BESS can be a temporary solution. Moreover, they must capture long-term development in the need triggering a grid planning measure (e.g. growth in load demand or PV). Finally, the methods for Cost-Benefit Analysis need a realistic modelling of the operational benefits of BESS. This comprises models that captures multi-period AC power flow with acceptable computation times, representative variations of load and generation over the year, realistic modelling of lifetime of the batteries, and last but not least multiple services and the trade-off between these services.
With this literature review as a foundation , it is now possible in CINELDI to develop Cost-Benefit Analysis that are in the research front, while focusing on the economic aspects of energy storage that are crucial for grid developers
Grid flexibility categorisation and modelling
Electric vehicles (EVs) in Norway and the potential for demand response
Work has been performed focusing on the consequences of the increasing share of electric vehicles and the potential for demand response and flexibility in charging. Results are based on a survey performed among households with electric vehicles and meter data of the energy consumption from charging of a selection of the most common electrical vehicles in Norway.
According to the survey most of the charging was performed in the afternoon and during the night, and approx. 50% of the households charge their EV from a normal socket (10 A). To map the potential for flexibility in time of charging, the respondents were asked about their willingness to postpone the time of charging from day/afternoon to night (hour 21-05). The respondents further supported the idea that if this shift in charging time had no negative consequences for the user, 90% were willing to postpone the time of charging, but if this reduced the driving distance the next day to 80%, the share of positive respondents was reduced to 56,5%. 38,2% of the respondents were positive to this change in time of charging if they save 200 €/year. A lesser amount (26,4%) of the respondents were positive if the savings are reduced to 50 €/year.
Households' potential for flexibility
A survey among a representative sample of households has been performed, evaluating the households' potential for flexibility. According to this survey 3 out of 4 households will delay the start of their washing machine and dishwasher until later the same day, on a cold day when there is limited grid capacity available. 2 out of 3 households can accept remote control of their water heater (as long as they do not get cold water) on such a cold day. Every second household will reduce their electricity consumption if this can help others to get electricity back after an outage.
Data-driven Household Load Flexibility Modelling: Shiftable Atomic Loads
This work describes a flexibility modelling method for atomic loads, which is based on high resolution appliance measurement data. Shiftable atomic loads are loads that can be shifted but once they start, they cannot be interrupted (cloth washing machines, dryers and dishwashers).
The practice of shifting the load from one hour to another is not simply cutting the load of the previous hour and adding it to the new hour. Rather, especially for shiftable atomic loads, flexibility modelling requires a careful study of both the consumption profile of the individual loads to be shifted and the temporal probability of use profile of the appliance in stock of households. Atomic loads cannot be interrupted and hence only the starting time can be shifted. This method can be used to quantify the flexibility potential of shiftable non-interruptible appliances.
The presented method will inform stakeholders how much power (kW) they can reduce by shifting the potential operation of the appliances. The activation of such flexibility resources, on the other hand, requires its own in-depth investigation as it may depend on the availability of communication channels to both customers and appliances, the willingness of customers, the market arrangement and the smartness of the appliances. Also, smart activation of the resources shall be executed to avoid rebound effects; for example, by distributing the shifting of a group of appliances over time instead of executing all resources at once. The rebound effect mainly arises from an increase in consumption due to the superimposing of shifted appliances on top of the already operating appliances. Hence, rather than spreading the shifting of the appliances overtime, one has to observe the probabilities of start operation for the next hours to decide the appropriate hour to shift to.
Analysis of Future Loading Scenarios in a Norwegian LV Network
The aim of this analysis is to support the realistic understanding of the potential network loading and power quality related problems coming in the network and to what extent flexibility resources could reduce or eliminate those challenges. An LV test network supplying residential area from 22 kV/ 230V secondary substation in Steinkjer in central Norway is used for the analysis.
Sizing electric battery storage system for prosumer villas
A simple energy flow model is developed where the battery is utilised to reduce the peak load from the villas (Skarpnes). The analysis show that there is a significant reduction in the daily peak load distribution by the use of battery storage systems. However, the peak-reduction effect is levelling out for larger storage capacities as measured relative to the consumption level of the villa.
Customers' involvement in flexibility
There is an ongoing PhD work with the objective to improve the understanding of how different grid customer groups can and will contribute with flexibility, and how DSOs and other stakeholders - especially retailers and aggregators - can utilise this knowledge to improve their services and profitability. Main research activities include in-depth studies on mechanisms and incentives for motivating customers to contribute with flexibility; analysis of the role of intermediaries which must be incentivised to facilitate or take up Demand Response (DR) services, contract design, and the related technological infrastructures and DR technologies to support the commercial products and new business models demonstrating the profitability of DR in the future power market.
A pre-study has been performed including interviews with 15 stakeholders from the industry, research etc., to study the definition of the term " flexibility mechanisms", and what they expect of future development within this area. This has been followed up with household interviews, where some have advanced flexibility systems installed (Stavanger), and a larger group has no such systems (Trøndelag + indre Østlandet). These interviews will generate data for further PhD work.
Value of flexibility
The impact of flexible resources on the security of electricity supply
A literature survey on the impact of flexible resources on the security of electricity supply (SoS) has been performed. The flexible resources that are considered are the following distributed energy resources: demand response (DR), stationary distributed energy storage systems (ESS), electrical vehicles (EV), and distributed (primarily photovoltaic, or PV) generation. Microgrids (MGs) are also considered to some extent. The survey distinguished between four aspects of security of electricity supply: energy availability; power capacity; reliability of supply; and power quality, which includes the voltage root mean square average value, the voltage waveform, and frequency quality.
In general, the requirements for services from flexible resources must be seen in relation to the time scales of the SoS issues they are addressing. Frequency regulation services to improve frequency quality operate on different time scales, and most flexible resources are capable of providing some of these services. But for fast frequency reserves to avoid rapid drops in system frequency due to loss of power injections, battery energy storage and demand response for certain loads are believed to have the greatest potential due to their short response time. All flexible resources except for demand response can provide voltage control. Other examples of potential services for improving power quality include phase balancing and damping of harmonics.
To pursue this goal, the work package is organised into four main Tasks:
- Grid flexibility categorisation and modelling
- Methods for sizing and placement of flexible resources
- Customers' involvement in flexibility
- Value of flexibility.
The major activities in the start of the year was to plan the research activities in detail, to recruit PhD candidates and to coordinate with the other CINELDI work packages as well as the relevant work within FME ZEN and FME CenSes. Moreover, the main achievements in 2017 are briefly described below.
Partner workshops on flexibility
At the 7th of September, we arranged a partner workshop together with WP3 and WP4, with focus on establishing good contact between the researchers and user partners, as well as discussing which technologies and solutions are most relevant to study in a 2030-2040 perspective. On the follow-up event during the CINELDI days 31st of October to 1st of November, we established the foundation for 3 pilots on flexibility, together with the user partners:
- Pilot 1: "Flexibility used in system operation"
- Pilot 2: "Flexibility as alternative to distribution grid reinforcements"
- Pilot 3: "Green and flexible end-users".
These pilots will be important assets for testing out the usefulness and relevance of new simulation models, planning methods business models etc. that will be developed as part of the research.
Energy Modelling workshop and 5th Pyomofest
Together with our international research partner Sandia National Labs, we arranged this workshop 3rd -5th of October in Trondheim, with around 50-60 national and international participants from academia, research and private sector. The main objective for the workshop was to gather scholars, researchers and analysts working with mathematical modelling and optimization of energy systems such as energy storage optimization and EV integration, with a special focus on the open-source programming languages Python and Pyomo.
Master theses on flexibility
Well-planned master theses can be key to establish a successful link between research and education. Regarding the FME's, the master theses serves more important purposes, as the user partners also can be actively involved in the process. We decided to "kick-start" the research by establishing 14 master theses at Dept. of Electric Power Engineering (NTNU) that are connected to the WP5 research for the years 2017 and 2018.
Other research highlights from 2017
- Successful power-hardware-in-the-loop tests of EV power grid interface in the SmartGrid lab (DTU-SINTEF-NTNU collaboration).
- Development of dynamic AC-power flow model with storage and EVs (Matlab model).
- Abstract accepted for special issue of "Sosiologisk Tidsskrift". Working title: det fleksible mennesket 2.0.
- Developed optimization models for prosumers (with PV, EV and storage) based on stochastic and deterministic dynamic programming (Matlab-models).
- Assessments of grid tariff structures for prosumers (MSc theses at NTNU).
- Assessments of how energy storage can exercise market power in small power systems (MSC thesis at NTNU).
- Three scientific papers were presented at international conferences:
- "Stochastic Optimization of PV Battery System Operation Strategy under different Utility Tariff Structures". Int. Workshop on Integration of Solar Power into Power Systems 2017.
- "Guidelines for DSOs on Reactive Power Provision by Electric Vehicles in Low Voltage Grids". CIRED 2017.
- "Integration of PEV and PV in Norway Using Multi-Period ACOPF — Case Study". IEEE PowerTech 2017.