Thermal Study of Valve Regulated Lead Acid Batteries and

Electronics Chamber Used in Stand-Alone Street Lighting

Applications
J. Skaalum, T. Jamieson, D. Groulx*
Mechanical Engineering, Dalhousie University, Halifax, NS, Canada
*Corresponding author: P.O. Box 15 000, Halifax, Nova Scotia, Canada, B3H 4R2, [email protected]

Abstract: This paper presents a study on the One area where the VRLA batteries are
heat generation of Valve-Regulated Lead Acid particularly vulnerable is temperature
(VRLA) batteries used in certain off-grid street- management. In general, when a lead acid
lighting applications from PoleCo, a Halifax battery receives a float charge at a temperature
based company. One goal of the project was to 10°C above its intended operating temperature,
produce validated COMSOL models of the the battery’s life will be cut in half [1]. This
enclosure that holds these VRLA batteries. This temperature effect is especially troublesome for
model can then be used to investigate methods of the PoleCo streetlights, because many of them
reducing the temperature of the batteries based are installed in locations with very high daytime
on environmental conditions. In order to provide ambient temperatures, such as Jamaica and
this validation, COMSOL models of experiments Qatar. In addition, the batteries themselves
were built and the results were compared to the generate some heat due to the effects of internal
associated experimental data. For models resistance to current flow as well as the
involving heat conduction in solids, the electrolyte chemistry within the cells [1].
COMSOL models provided very accurate In order to develop a better understanding of
results. For models incorporating natural the heat generation characteristics of their street-
convection and the influence of additional lighting system, PoleCo is collaborating with the
electrical components, producing accurate Lab of Applied Multiphase Thermal Engineering
COMSOL models proved more challenging. (LAMTE) at Dalhousie University. Over the
course of this project, experiments on the VRLA
Keywords: Heat Generation, Battery Thermal batteries and associated equipment have been
Management, Conduction, Natural Convection. performed. One phase of this project is to also
create and validate numerical models of the
1. Introduction experiments using COMSOL Multiphysics 4.4.
This will lead to the next stage of the project,
PoleCo is a Halifax, Nova Scotia based using the validated numerical models of the
company that designs and supplies grid- system to test methods of mitigating the heat
independent street lighting systems. These street gain in that system, such as through solar
lights derive all of their power from renewable shielding or active cooling via fans.
sources using small wind turbines, solar panels, This paper will present a brief summary of
or a combination of the two. In order to store the the experimental work carried out during this
power, each street light has on-board project, and then describe in detail the COMSOL
rechargeable Valve Regulated Lead Acid models designed to simulate the experiments.
(VRLA) batteries. These batteries store power The results of these simulations will be
generated, for example, by the solar panel during compared with the associated experiments both
the day so that the streetlight can reliably draw to validate the models and to test the accuracy of
power all night. The batteries are selected for calculations aimed at predicting temperature rise
their durability and long cycle lives, so that they of the batteries.
can operate continuously for long durations in
the field. Replacing these batteries is expensive 2. Use of COMSOL Multiphysics
both because of their unit cost, and also the 2.1 Insulated Battery Experiment
expense of sending repair crews into the remote 2.1.1 Geometry and Materials
off-grid locations where the streetlights are
usually installed. Hence, PoleCo has great The first experiments performed on the
interest in methods for prolonging the lifetime of battery were designed to measure its temperature
the batteries. gain when charged at different rates. The

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

experimental setup involved one VRLA battery
enclosed in a tight-fitting box made from half-
inch RSI-0.5 extruded polystyrene (XPS). The
battery was charged with a bench-top DC power
supply set to 14.5 V and various charging
currents. The temperature of the battery was
measured with fourteen adhesive T-type
thermocouples arranged along the surface of the
polypropylene casing.
In COMSOL, this experiment was modeled
as two blocks, one representing the battery and
the other representing the insulation, as shown in
Fig. 1. The battery domain is 287 mm tall, 342
mm long, and 172 mm wide. The insulation Figure 1. COMSOL Geometry of Insulated Battery.
block is subdivided into six panels to make it
easier to mesh. This geometry is extremely
simplified, and does not include any of the
surface features of the battery such as handholds
and electrical terminals, as seen in Fig. 2. In
addition, the internal geometries of the cells that
make up the battery were not included in the
COMSOL model.
The thermophysical properties of the
materials used in this simulation are defined by
the values in Table 1.

Table 1: Material Properties used in

COMSOL Simulation
Material Insulation Battery Figure 2. Photo of Insulated Battery with
Thermal Conductivity 0.0254 34 Associated Experimental Equipment.
(W/m·K)
Density (kg/m3) 21 [2] 2841
Heat Capacity (J/kg·K) 1500 [3,4] 862.3

The density of the battery was calculated

from direct measurement of its mass and volume.
The heat capacity and thermal conductivity were
calculated using a weighted average of the
materials inside the battery, taken from the
material data safety sheet [5].
In addition, the battery domain has fourteen
boundary point probes arranged along its surface
to record the temperature. These point probes are
placed in locations that match the positions of
the thermocouples used in the experiments to
measure the battery temperature, as seen in Fig. Figure 3. Schematic of Thermocouple Positions on
3. The purpose of these probes is to provide a Battery.
convenient method for comparing the
temperature data from the COMSOL simulation 2.1.2 Governing Equations and Boundary
with the temperature data collected Conditions
experimentally.
The Heat Transfer in Solids module is
applied to all the domains. The initial

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

temperature of the battery and insulation
domains is set to a representative value from the
charge experiment that is to be validated. For
example, in Charge Experiment #5 (CE #5) the
initial temperature was 22.2°C. The faces of the
insulation not facing the ground are given
boundary conditions appropriate for natural
convection and radiation to ambient conditions at
22.2°C; the bottom surface in contact with the
workbench is fixed at 22.2°C.
The battery domain is assigned as a heat
source with power defined by a function specific
to the experiment. The heat generation due to
Figure 5. View of Meshed Geometry.
internal resistance inside the battery is given by
Eq. (1) [6]:
2.1.3 Meshing and Mesh Analysis
POV = IDV (1)
As the model geometry is formed from
rectangular elements, it was possible to mesh it
where POV is the heat generated (W), I is the
using swept tetragonals, as shown in Fig. 5.
current flowing through the battery (A), and DV Tetragonal mesh elements were chosen to
is the over-voltage (V). Over-voltage is the improve the convergence characteristics of the
difference between the open circuit voltage of model, as well as save computational power by
the battery before charging begins and the limiting the number of elements.
measured voltage at the battery terminals during In order to determine the appropriate size of
charging. The values of I and DV for an the tetragonal elements, a simple mesh analysis
experiment can be determined from measured was performed. The simulation was run with
values and are presented in Fig. 4. mesh sizes corresponding to eight of the
The data presented in Fig. 4 is then predefined settings in COMSOL. The
simplified into a series of linear functions that temperature reading from boundary point probe
are plotted in a piecewise fashion and set as a #1 at the final time step was recorded for each
uniform power generation for the battery run. Figure 6 presents that temperature as a
domain. function of the mesh size.

Figure 6. Comparison of Probe #1 Temperatures for

Figure 4. Power Associated with Over Voltage for
Varying Mesh Sizes.
Charge Experiment #5.

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

 Overall, the difference in temperature
between the finest and coarsest settings used is
less than 0.004 °C. In addition, the difference
between the Normal mesh setting and the finer
meshes is extremely small compared to the
differences between the coarser settings. Based
on the trend presented in Fig. 6, the Normal
predefined mesh setting was used. The complete
mesh has 3553 elements. At this mesh size, the
simulation is able to complete in less than 5
minutes.

2.2 PoleCo Enclosure Experiment

2.2.1 Geometry and Materials Figure 8. View of COMSOL Model of PoleCo
Enclosure Experiment.
When the VRLA batteries are in use on a
PoleCo lighting installation, they are held within To model the flow of heat through solids and
an 18 inch square aluminum enclosure at the top fluids, as well as natural convection effects, the
of the light-post. Later experiments performed on Conjugate Heat Transfer with Laminar Flow
the battery focused on how heat moved around physics was used. This model is defined by four
within this enclosure, as shown by the domains; the aluminum box, the air inside the
experimental setup in Fig. 7. box, the two batteries, and the charge controller,
The air cavities within this chamber are large as shown in Fig. 8. Of these, the box and the air
enough that convective currents are expected to use the standard material properties from the
have a noticeable effect on the temperature of the COMSOL library. The batteries use the same
batteries. There is a 40 mm gap between the material properties as the ones in the model of
batteries and a 65 mm gap between the enclosure the insulated battery experiment.
walls and the batteries on the four sides. There is Investigation of the charge controller,
150 mm of space above the two batteries. The however, showed that it needed to be treated
charge controller has a 128 mm square base and with special attention. In a street-light
is 73 mm tall. To better isolate the effects of installation, the charge controller modulates the
these air currents, the exterior of the enclosure current and voltage characteristics of the
was completely wrapped in polystyrene electrical energy from the solar panel. The
insulation, similar to the insulated battery purpose of this modulation is to ensure that the
experiment. battery is always being charged at the highest
possible power regardless of the position of the
sun. As a result of this operation, the electronic
circuitry within the charge controller generates a
noticeable amount of heat. For the purposes of
validating numerical models of the experiment,
the charge controller has been modeled as a mass
with a predefined temperature, set to match the
experimental results from charge experiment #15
(CE #15). CE #15 was the first charge
experiment performed in the insulated PoleCo
enclosure.
For this simulation, the batteries are acting as
heat sinks for the charge controller, in addition to
generating heat themselves. As heat is flowing
into the battery from an external source, it is
important that the battery domain in this model
Figure 7. Photo of Batteries and Charge Controller accurately represents the internal structure of the
within PoleCo Enclosure. actual battery. In the top section of the battery,

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

for example, there is an air gap that contains
gasses released by the battery during operation.
This air gap acts as an insulator, slowing heat
flow from the roof of the battery downwards. For
the purposes of modeling, this gap was estimated
to be 50 mm tall.

2.2.2 Boundary Conditions and Governing

Equations

Similar to the insulated battery experiment,

the batteries are defined as uniform masses, with
Figure 9. Meshed Geometry for PoleCo Enclosure
the exception of the 50 mm air gap on the top
Model
surface. The heat generation is given by a
function taken from experimental data. The
addition of a fluid domain automatically defines
wall domains on the interior of the enclosure and
on the surfaces of the batteries and charge
controller. All air flow is due to the natural
convection, thus slowly flowing in a laminar
manner with no slip condition at the walls. The
outside faces of the box are thermally insulated,
representing the polystyrene. A buoyancy force
is included in the model as a volumetric force
given by:
FZ = -g*(rho-rho_ref) (2)
Where g is the gravitational constant, rho is the
density of the air, and rho_ref is the density of
air at 1 atm and 25 °C. Figure 10. Temperature Plot of Battery Cross Section

2.2.2 Meshing

Figure 9 shows the mesh used for this model,

which was also a swept tetragonal due to the
rectangular nature of the geometry. Several
parametric surfaces were used to divide up the
geometry in order to facilitate this. The mesh
shown below has 5760 elements, and the
simulation completes in a time of 30 to 40
minutes.
3. Experimental Results
3.1 Insulated Battery Experiment

This simulation is designed to be compared

with an experiment than ran for 200 minutes.
Thus, the transient solver was used to generate Figure 11. Comparison of Temperature Data from
all the results. Figure 10 below shows the results COMSOL Probes with Experimental Data from
from a simulation using the conditions associated Charge Experiment #5
with charge experiment #5 (14.5 V, 18.5 A) at
the 200 minute mark. The results of the probe
comparisons can be seen in the Fig. 11.

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

 Because the battery was represented by a
single uniform domain, there is only a negligible
difference in the temperature readings between
boundary point probes. Hence they are all
represented by a single line. By the end of the
simulation the difference between the
experimental data and the probe data from the
COMSOL model is relatively small. The
majority of the thermocouple temperatures are
within 0.5°C of the simulated results, while the
rest are within 1°C. Note that the standard
tolerance of the type-T thermocouples used for
the experiment is 0.5 °C [7].
Figure 13. Comparison of Battery Temperature Data
The close match between the COMSOL from COMSOL Probes with Experimental Data from
simulation and the experimental results suggests Charge Experiment #15
that using the over voltage measurements is a
reasonably accurate method of calculating heat
gain in the battery.

3.2 PoleCo Enclosure Experiment

Figure 12 shows a series of cross sections of

the solution provided by the COMSOL model at
the end of a 500 minute transient simulation.
In this solution, every element within the
enclosure is at approximately the same
temperature, slightly warmer near the ceiling by
about 0.6 °C. At this point in the simulation the
air has developed stratified layers due to the
effects of natural convection. Figure 14. Comparison of Enclosure Temperature
Data from COMSOL Probes with Experimental Data
from Charge Experiment #15

Figure 13 shows a comparison between the

data collected from the COMSOL probes on the
battery domains and the experimental results.
Overall the temperature trend seen in the
experiment is matched by the simulation,
although the temperatures are slightly under
predicted. Notably, however, the addition of the
air gap on the top of the battery helps capture the
sharp rise in temperature seen on the roof of the
battery. Note that radiation heat transfer within
the enclosure was not simulated at this time.
Figure 14 shows the results from the
COMSOL probes associated with the enclosure,
compared with experimental data. The trend
shown by the COMSOL probes does not match
the experimental results precisely, although the
magnitudes of the temperatures are very close. It
Figure 12. Temperature Plot of Enclosure at Various is likely that modeling the internal structure of
Cross Sections
the battery more accurately would result in a

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston

simulation that more closely matches the 4. H. Winterling, N. Sonntag, Kunststoffe
experimental data. International, 10, pp. 18-21 (2011)
5. Concorde, Concorde Battery Valve Regulated
4. Conclusions Lead Acid Battery, [MSDS], (2014), Retrieved
from:
For the purposes of producing useful http://www.sunxtender.com/otherpdf/vrbatts_ms
numerical models of the PoleCo battery ds.pdf
enclosure system, two types of COMSOL
models were constructed based on experimental 6. H. Giess, Investigation of Thermal
setups. Phenomena in VRLA/AGM Stationary
The first type of model simulated Lead/Acid Batteries with a Thermal Video
experiments performed on a single battery Imaging System, Journal of Power Sources, 67,
wrapped in insulation. In this model the system 49-59 (1997)
was simplified greatly, essentially breaking it 7. ASTM Standard E230, “Standard
down to a single uniform mass with certain Specification and Temperature-Electromotive
material properties. By applying heat generation Force (emf) Tables for Standardized
equations based on the over voltage applied to Thermocouples”, 2012, www.astm.org
the battery, it was possible to simulate numerical
results that were comparable to observed 6. Acknowledgements
temperature changes from experiments. In this
manner, the use of COMSOL helped to confirm The authors would like to thank PoleCo for
that for the purposes of heat generation, the their time and collaboration, as well as the
battery could be treated as a very simple object. Natural Science and Engineering Research
The second type of model was designed to Council of Canada (NSERC), the NSERC
mimic an experimental setup involving two CREATE funded Dalhousie Research in Energy,
batteries inside a small box, along with a charge Advanced Materials and Sustainability
controller. Unlike the previous model, it was (DREAMS) program and the Canadian
important to accurately model the interior Foundation for Innovation (CFI) for their
structure of the battery so that the transfer of heat financial contribution to this work.
through it from an external source would be
modeled correctly. The results from this model
approximately match the experimental results,
and could be refined further in the future to
provide better results.
Overall the COMSOL models helped show
that using over voltage as a measure of the heat
generated within the lead-acid battery during
charging was a valid assumption.

5. References
1. D. Berndt, Valve-regulated Lead-acid
Batteries, Journal of Power Sources, 100, 29-46
(2001)
2. ASTM Standard C578-14, “Standard
Specification for Rigid, Cellular Polystyrene
Thermal Insulation”, 2014, www.astm.org
3. K. T. Yucel, C. Basyigit, C. Ozel, Thermal
Insulation Properties of Expanded Polystyrene as
Construction and Insulation Materials, 15th
Symposium in Thermophysical Properties, pp.
54-66 (2003)

Excerpt from the Proceedings of the 2014 COMSOL Conference in Boston