The Rholab Project has successfully demonstrated that a lead-acid battery can power a Honda Insight. Now the Advanced Lead-Acid Battery Consortium has a new UK Foresight Vehicle project which seeks to develop this technology further for lower voltage application as Pat Moseley and Allan Cooper explain.
It has been extensively reported that, if batteries continue to be the energy storage devices of choice in future automobile electrical systems, they will need to supply and to accept charge at unprecedented rates.
For example, in the ALABC’s Rholab Project it was shown that the Honda Insight frequently draws 100A from its 6.5Ah battery during power-assist. Thus the battery is required to support discharge events at around 15 times the 1-hour rate and, when regenerative braking is deployed, recharge will be required at about half this rate.
It is also now well established that when a lead-acid battery is operated at such rates and is not regularly brought to a full state-of-charge, i.e. when it is operated according to a high-rate partial state-of-charge (HRPSoC) regime, as in a hybrid electric vehicle, a novel failure mode appears. Whereas traditional automotive batteries generally ultimately fail due to processes occurring at the positive plate (grid corrosion, active material shedding), the battery in HRPSoC fails as a result of the accumulation of sulfate on the negative plate, principally at the outer surfaces.
A clear demonstration of the change in mechanism was provided in the Exide Technologies paper presented by Maria-Luisa Soria at 9ELBC in Berlin. In deep discharge cycling at the C/10 rate and in partial state-of-charge (PSoC) duty at a moderate rate (C/3), a 15 Ah VRLA cell exhibits the symptoms of failure according to Type-2 premature capacity loss (PCL-2), an expansion of the positive active material. There is also some sulfate towards the bottom of the negative plates, indicating acid stratification. In higher-rate PSoC operation, however, the major change in the cell is a massive accumulation of sulfate at the top of the negative plate.
It is evident that, at the highest rates, charge is principally drawn from, and restored to, the region closest to the current take-off tabs. The active material near the tab is worked very hard while the more remote material, further down the potential gradient from the tab, is hardly used (see Figure 1).
An earlier article on the Rholab project (see BEST, Summer 2004) described how the uneven use of the active material across the plate can be ameliorated by the use of a twin tab approach. When a second tab is attached to each grid opposite the first, a far greater fraction of the active material experiences the full potential environment and takes part in discharge and charge reactions. As a result of this approach a spiral-wound cell built during this project, with take-off tabs at top and bottom of the plates, was able to manage a heavy HRPSoC duty repeatedly for long periods (Figure 2) while a similar cell without the second tab was unable to support the cycle even once (see Figure 3).
The success of the twin-tab design in this demonstration leads us to ask if equivalent improved operation in HRPSoC conditions might be achieved with other grid designs. Clearly such designs should be based on minimising the potential drop between the electrical connection to the external circuit and the most remote component of the active mass. This is a problem which can be tackled effectively by a modelling approach. ALABC work carried out by the team at the Technical University of Brno in the Czech Republic has provided evidence that the grid design shown in Figure 4 goes a long way towards evening out the current distribution over the electrode area.
A follow-up project to the Rholab work in the UK Foresight Vehicle Programme, which again has funding from the ALABC as well as the Department of Trade and Industry and the Engineering and Physical Sciences Research Council, has recently started. This is known as ISOLAB42 (the acronym for Installation and Safety Optimised Lead Acid Battery for 42V Applications). The intention is to develop further the technology from the Rholab work while perhaps simplifying it to make it suitable for 36V mild hybrid vehicles. The twin tab approach, though highly successful in achieving high-rate performance, does complicate the construction and electronic monitoring of the battery. Thus alternative approaches to both flat-plate and spiral-wound designs are being looked at by the two battery partners – FIAMM and Exide, respectively.
Taking the FIAMM design first, the grid starts as an extrusion utilising the Teck Cominco process and is designed to optimise the current collection from the grid. This has a wide selvage running the full height of the plate and all the positive and negative plates will be in contact with each other at the sides of the plates. The extruded dimensions are very precisely controlled by the die, and the grids are subsequently expanded sideways. The positive grid has a thickness of 1.6mm, with the negative thickness being 1.4mm. The design is shown in Figure 5.
The separator will be AGM, zig-zag wrapped to include the end plate. In order to obtain even compression of the group a plastic spacer will be used at each end of the element stack, and it may have a polypropylene band to ensure full compression. This would also facilitate handling while the group bar is welded. Group bar welding will be simplified as it will only be necessary to fuse the tops of the tabs together with adequate penetration and join them to the terminal pillar.
The element group will have seven positive plates and eight negatives and a ratio between positive and negative active mass to achieve a compromise between performance and cyclability. FIAMM experience with VRLA automotive batteries for normal SLI and stop-and-go applications has been incorporated into the design. The ABS case has features on the outside to enable the 2V cells to be made up into batteries of different configurations, as will be described later.
While the low rate capacity of this design is anticipated to be in the order of 30Ah, it is hoped that it will also be possible to investigate this type of cell in lower capacity format, i.e. closer to the 8Ah used in the Rholab project. This could include the conventional single tab format, a design with positive and negative terminals at opposite ends of the cell, or even a dual-tab design in order to maximise discharge/recharge ability.
The Exide work is different in that the design is based on its Orbital spiral-wound battery, but this will be developed and optimised for HEV applications. It is being produced as 6V modules to fit in with the concept of being able to design batteries to fit into the most appropriate location in the vehicle.
While adjustments have been made to both positive and negative paste, the principal design changes have entailed the incorporation of through-the-partition connection between the cells to ensure the isolation (and consequently improved reliability) of the individual cells, as well a larger acid reservoir to improve the filling and formation of the batteries. This larger top space, together with the use of a high pressure Bunsen valve, also provides for a larger gas accumulation to reduce water loss in the cells during the HRPSoC service. Finally, a new flat plate female terminal has been designed to allow faster, easier and more flexible connectivity to be achieved. Cooling access has also been improved.
Figures 6 and 7 show a battery in conventional 36V format in part-assembled and completed form. This may not, however, represent the best option for fitting the battery in the vehicle.
As the ISOLAB42 name suggests, one of the central themes of the project is installation optimisation. Vehicle electrical loads are continuing to increase, particularly in the premium sector, and peak loads may now exceed the maximum power output of existing alternators. The situation is exacerbated with hybrid powertrains, in which the batteries are also required to provide motive power; consequently the current trend towards higher capacity, and hence larger, batteries is likely to continue.
However an increase in battery volume is likely to be very unpopular. Packaging of the conventional SLI battery can already pose a significant problem for vehicle manufacturers, and we are reaching the point where locating the battery as a large monolithic block under the bonnet is no longer viable. In order to accommodate ever larger batteries it may well be necessary to alter our views on conventional battery design and even break the battery up into smaller units and distribute these into available locations around the vehicle.
This naturally leads to questions of compromise concerning issues such as the loss of useful space, mass distribution, interconnect and thermal management. The Warwick Manufacturing Group (WMG) of the University of Warwick is building upon previous packaging experience with the Rholab Project (and over 20 years of automotive manufacturing research) to identify the optimum location and configuration for the battery.
One of the main difficulties to overcome is the level of compromise acceptable to the car manufacturers. Packaging space is at a premium in all modern vehicles and there is no obvious location that can be identified without at least an element of redesign, even taking into consideration that modern sealed cells can be mounted in any orientation.
Figures 6 and 7 show the Exide modules configured into a 36V battery with the 6V modules in a vertical orientation. An approach that accepts that some redesign will be necessary, combined with the possibility of distributing the battery leads to a much greater number of options. WMG’s research aims to identify the issues that arise from the introduction of these novel packaging options, and to quantify the trade-offs to allow manufacturers to make better-informed decisions.
In addition to the physical space required to package the cells, another key consideration is thermal management. It is well documented that service life of any battery depends upon operating temperature, and it is also known that non-uniform temperature distribution across battery cells also leads to significant degradation of useful life. This is particularly relevant to mild hybrid applications, which operate over a much harsher duty cycle than SLI batteries. Temperature differences between cells could lead to differing state-of-charge and could make effective battery management very difficult to achieve.
The aim is to make this installation research generic. It is hoped that it will be possible to identify suitable locations for cell or monobloc level installation of a potentially distributed battery into any vehicle. The approach that has been adopted is to develop the CAD solid models of the battery components, develop rules concerning the compromises necessary, and then combine these into a number of concept designs.
Within the lifetime of the project the method will be expanded to interface to CAD data for the vehicle, thereby allowing the method to be applied to any vehicle which has a suitable CAD model. Which compromises to accept then falls to engineering and marketing judgment, but these decisions can be taken in the light of much better information than traditionally available.
The ISOLAB42 project aims to validate its research on a demonstrator vehicle. Three promising concept designs have been selected: spare wheel well, distribution around the luggage compartment, and under seat. These have been developed into dimensioned models with fixings and enclosures to prove the concept is viable and allow thermal analysis and airflow modelling to proceed. Figure 9 shows how the spare wheel well concept could work with the FIAMM cells. This option has been selected because there seems to be less reliance on the spare wheel as a ‘get you home’ feature on modern vehicles.
WMG will continue to develop an understanding of the issues and trade-offs involved in packaging a battery, undertake detailed mechanical design (including ducting and airflow requirements), and finally realise prototype enclosures for the demonstrator vehicle using rapid prototyping and tooling. This technique was applied successfully in the Rholab project, and can generate serviceable components in a matter of hours rather than weeks without the need for complex and costly tooling.
As with the Rholab project, the battery management system (BMS) is being developed by power electronics expert Provector. The higher currents expected in the ISOLAB battery have precluded the use of printed circuit board (PCB) tracks to carry the main battery current. Nevertheless there are still a number of advantages to using a module-mounted PCB to carry the BMS components, and this is part of the ISOLAB design requirement.
However, because the final topology of the system will not be finalised until later in the project, it was advantageous to construct an interim BMS system on separate, modular hardware. This allows the system software to be written and tested prior to the battery modules being built, with immediate transfer to the new battery modules as soon as they are available. It would then be a relatively risk-free step to construct customised module-mounted hardware using the proven BMS elements to suit the final battery system design. Because the modular hardware was less restricted by space than the final version, it was decided to add the capability to use a pressure transducer and reference electrode on each cell as required during development work. This had the benefit of tightly integrating these results with the other measurements.
Experience from using the Rholab BMS has led to the incorporation of a number of changes to the design for ISOLAB:
• The internal network was changed to a CAN implementation to provide hardware error correction and avoid translation between the internal and external formats. This also allows larger physical separations between the various modules where the battery pack is distributed within the vehicle as envisaged in the project. LIN was also considered as a second automotive standard, but had an insufficient data rate for the application.
• The microcontroller used for cell monitoring was changed to reduce quiescent power consumption, upgrade the analogue inputs to 12-bit resolution, to provide on-chip CAN support and to provide greater code space, which had been restricted on Rholab.
• The cell monitor electronics (Figure 10) are now powered directly from the cell through a small switch-mode power supply boosting or bucking to the 3V required by the microcontroller depending on whether the system is monitoring the FIAMM 2V cell or the Exide 6V module. The power supply is specified to operate down to a cell voltage of 0.8V with the FIAMM cell and 2.4V with the Exide. Current drawn by the cell monitor is also monitored as part of the energy audit system.
• A single PCB is used to provide 2V, 6V and 12/24V versions through selective assembly.
• The conditioning system has been separated into a separate power module and some support components in each cell monitor module associated with switching the conditioning supply on to the cell. This makes it easier to build systems with or without cell/module conditioning. Each power module is capable of around 20W power delivery during conditioning.
• The system is designed to support up to four conditioning power modules and up to 250 cell monitors to give the system stretch potential to include hybrids.
• Control of the conditioning voltage or current is much more versatile, and conditioning current is measured separately to high resolution to assist in determining end points and assess overall energy flows.
• The software has provision for insertion of algorithms being developed by the University of Sheffield as part of the project.
The University of Sheffield will be responsible for some of the battery testing during the project. However, pending the completion of development work on the batteries, models based on empirically determined cell characteristics are under development.
The models will be used within the ISOLBAB42 battery management system to aid in the estimation of cell state-of-charge (SoC) and state-of-health (SoH) and hence provide a cell state-of-function (SoF).
The approach has employed both classical equivalent circuit modelling and the application of Kalman Filters (KF) and Extended Kalman Filters (EKF) to state space models of the cells.
State space modelling techniques should provide a robust solution to the problems of modelling the complex, non-linear parameters found in battery systems, as the state space model gives a description of the internal and external characteristics of a system.
The states describing the system do not necessarily have physical meaning, but do contain historical information on the system.
The use of KF offers a method for estimation and recursive correction of the states of the model. This can be carried out efficiently in the presence of additive noise signals, as occur on the bus bars of battery packs within hybrid electric vehicles.
The state space battery models are being developed within a dSpace real time digital signal processing system connected to the University’s cell and pack testing benches.
Initial results from the state space approach are encouraging. In particular, a comparison of SoC estimates using the proposed KF technique, and the more conventional integration of current method, undertaken using ‘road data’ collected from the Rholab Honda Insight HEV driven on a test track, show a significant improvement in SoC estimation.
Furthermore extensions for SoH monitoring by employing EKF have also been undertaken, using only measurements of cell terminal quantities as input.
Such data is extremely important for the ultimate prediction of SoF, which describes the ability of a cell to perform adequately under HEV demands, and is related to both SoC and SoH information.
SoF will give a prediction of available capacity, and discharge and recharge capability, thereby allowing a ‘smart’ battery to forecast the response of the cell to driving demands.
This should lead to optimal utilisation of the battery pack with regard to performance and lifetime, and therefore better overall energy management within the vehicle.
Cell and module development is progressing well and some laboratory testing has begun. The control electronics will be ready for when bench testing starts in earnest at Sheffield. In the meantime discussions are in hand with vehicle manufacturers to discuss installation issues and to find suitable opportunities for demonstrating the developed battery systems.
