University of Sussex Mathematics School along with Innovate UK‘s Knowledge Transfer Network are running a 3-day Study Group on Mathematics in Energy Systems, between the 8th – 10th January 2018.

Overview

This three day session will be an intense problem solving activity inviting some of Britain’s best mathematical scientists tackling some of the hardest problems in the energy systems industry. We have attracted problems across the energy systems area – from storage, network supply, integration etc.

The importance of new and novel energy solutions is highlighted in recent investments by Her Majesty’s Government. The Faraday Challenge, for example, is an investment of £246 million over 4 years to help UK businesses seize the opportunities presented by the transition to a low carbon economy, to ensure the UK leads the world in the design, development and manufacture of batteries for the electrification of vehicles. This connects to many mathematically challenging topics including modelling degradation, power management, and integration.

Update:  The prospectus is now available.

The following industry participants will be presenting problems at the study group:

Williams Advanced Engineering

Electric Vehicle Degradation Study

Abstract:  Williams Advanced Engineering is the technology division of the Williams Formula 1 team. It provides world-class technical innovation, engineering, testing and manufacturing services to deliver energy-efficient performance to the motorsports, automotive, defence and healthcare sectors.  Williams have developed high power density batteries and electric motors for a range of applications and market sectors where high performance, endurance and predictability are key customer requirements. They have expertise in the design and manufacture of the Battery, Mo- tor Generator Unit, Inverter and Battery Management System.

Objectives: This challenge is comprised of three related sub-problems:

• Examining the Correlation Between Track Events and Battery Degradation: In performance electric vehicles, many of which may be used for non-standard driving such as track days, many different events may occur and much data can be collected. We wish to understand the rate of degradation as a function of different events (track characteristics, environmental conditions etc). Ultimately, we would like to develop an effective cost function for degrada- tion given the different nature of each cycle for individual cells and individual modules to each battery unit to its limit and also to inform the optimal design of the modules.

• Fast Charging of Batteries: After operation batteries are often warm when they are con- nected for recharge. To control degradation effects during fast charging application, cool- ing prior to or during recharge may be required. The challenge here is to derive a cost function and an optimal strategy for fast charging and cooling of car batteries. This could be done either combining existing empirical degradation models or starting from first prin- ciples.

• Cost Models for V2G applications: In V2G applications, the degrading effect of grid load balancing on individual vehicle batteries will be very important. There is the need for robust models which capture monetary costs to the effects of battery degradation due to this load management..

National Grid

Power system control – understanding the challenge of the UK‘s smart energy grid

Abstract:  The growth of renewable generation has resulted in a significant fall in power system inertia. National Grid needs to keep the inertia of the power system above a certain threshold to ensure that it can ride through the most severe disturbances, and this is now costing tens of millions of pounds a year. These costs are incurred because, at times of low system inertia, it is necessary to stop renewable generation and replace it with thermal generation (mostly gas-fired) which has a large spinning mass.

Objectives:  The growth of renewable generation has resulted in a significant fall in power system inertia. National Grid needs to keep the inertia of the power system above a certain threshold to ensure that it can ride through the most severe disturbances, and this is now costing tens of millions of pounds a year. These costs are incurred because, at times of low system inertia, it is necessary to stop renewable generation and replace it with thermal generation (mostly gas-fired) which has a large spinning mass.

At present there is no way to actually measure system inertia before a serious disturbance happens, so National Grid has to forecast it based on expected generation and demand conditions. These forecasts are used to decide which actions need to be taken to keep system inertia above the threshold.
National Grid believes that it may be possible to determine system inertia in real time based on analysis of the short-term variations in the 50Hz power system frequency. This exhibits continuous "noise" across a whole range of frequencies, and it is possible that the characteristics of this noise may provide information on the actual system inertia. Analysis of historic frequency data may also provide a range of other useful insights, so project objectives could include:

1. Spectral analysis (or other types of analysis) to determine the characteristics of the grid frequency “noise”.
2. Explore any correlations between this “noise” and National Grid’s historic forecasts of system inertia.
3. Analyse historic frequency data over several years to see how the behaviour of system frequency has changed with the growth in renewables.

Additional exploration of the data for other interesting phenomena could also potentially lead to:

1. the development of methods for automatically identify “interesting” events in historic data, so that historic and recent events can be compared;
2. determining if it is possible to identify differences in frequency management practices between different shift teams, which may help to identify best practice;
3. exploring whether there are noticeable regional differences between system frequency measured at different locations
4. exploring limited amounts of high-speed measurement data of system frequency (one measurement every 20ms compared to one every 1s or 0.1s in the main dataset) to see whether there are interesting phenomena in the data.

 Algorithms to Devise Optimal Power System Control Actions

Abstract: National Grid controls the power system in real time by issuing instruc- tions to power stations. However, these instructions have unique constraints on the form they can take, which are cur- rently solved by humans but which are very challenging for conventional optimisation algorithms to solve. For example, instructions must have a minimum amplitude and duration, must ramp up and down at prescribed rates, and cannot change ramping direction instantaneously. It is desirable not to issue instructions too early because of uncertainty on the power system, but if instructions are left too late then an unmanageable ”bow-wave” of instructions may be created. A rule-based solution to this problem has proved too complex, and mixed-integer linear program- ming is likely to be too slow (even if the problem could be formulated accurately).

Objectives: National Grid would like to investigate alternative algorithmic solutions which may be able to turn ”unconstrained” control actions (determined by a conventional optimiser) into practical instructions that meet the various hard and soft constraints described above.

Registration and Fees

The majority of the Study Group is paid for by the sponsoring organization (Innovate UK); however to cover some costs, there will be a nominal registration fee of £40. This fee will cover accommodation and meals for the three days. We also ask that travel costs should be sought from attendees’ own institutions.  To register please use the following link: Registration

Venue Details

The event will take place in the Fulton building at the University of Sussex (building C5 on the campus map).  The University of Sussex is located in Falmer which is a short bus or train ride away from Brighton (directions to the University).  Further local information is available at www.visitbrighton.com.

Funding

The meeting is supported by Innovate UK’s Knowledge Transfer Network, by the Department of Mathematics and the University of Sussex through the Sussex Research Impact Fund.