This web page is less and essay than a collection of photos, graphs and text describing
my 2003 Prius (model NHW11) battery experiments.
For now, consider it a journal or notes for a future chapter.
By happy accident, a salvage traction battery came available on Ebay:
Eventually I will instrument and refurbish this battery for advanced
testing. However, my current battery is quite healthy in part because
I take pains to keep it from over heating.
MPG After Battery Replacement
Patrick Wong had a warranty battery replacement in his former, 2001 Prius.
Earlier data from the Dept. of Energy fleet studies had suggested that
a new Prius has a slightly better mileage performance in the first 3-4 months.
Patrick's data suggests replacing the battery does boost mileage for a short period of
Battery Compression Plates
Unlike a standard, steel cylinder NiMH battery, the thin plastic sides
can not contain any significant pressure from gas formation.
This means we need compression plates with holes for the interlocking
plugs and nipples.
In my case, I used thin plastic sheets from a hobby shop to build the battery
But I built up layers of plastic and aluminum foil to make a capacitive, load
The capacitance increases as the batteries swell and push against the compression plates:
This shows a module under test using an MRC 989 battery charger and the capacitance
tester measuring the stress as the battery is charged and discharged.
Comparing NHW20 and NHW11 Modules
There were substantial improvements in the NHW20 modules over the NHW11.
The case visibly has more plastic and the terminals are slightly lower and
out suggesting the cronic leaks have been addressed.
These changes also mean we can not mix NHW20 modules with NHW11 but it
may be possible to replace all 38 NHW11 modules with NHW20s:
Unrecoverable Shorted Cell
In December 2007, another owner of a 2003 Prius decided to pay
for his out of warranty, battery replacement.
This made sense because the $4,500 was much less than trying to buy
a replacement, 45 MPG car.
I bought his old battery pack knowing the modules would be worn out
and I could use them to experiment with battery refurbishment.
Sure enough, of the 38 modules, one had an unrecoverable, shorted
cell. No amount of water replacement can correct a cell whose
separator has melted and shorted the electrodes:
I soon discovered that gas generation was pretty aggressive even using
the relatively low currents, 6 A., from my MRC 989.
In this photo, bubbles of electrolyte after water replacement are
This is a very caustic, KOH solution that will eat paper, cloth, eyes or other
It feels slipery but you do want to wash it off and keep it away from
cuts and open sores.
Inside of Module
Upon realizing this module had a permanent short, I decapitated it
to understand the internal structures and get parts for further
Alternate Sealing Mechanisms
Knowing these modules lose water due to electrolysis and heat,
I investigated using stainless steel "tee nuts" as water fill
Subsequent failures of the "tee nuts" suggest that one-sided, heat
sealing is not sufficient. It may be possible to put a plastic
cover over the top of the tee nut but different thermal expansion
rates make this desirable modification somewhat impractical ... darn it!
One by-product of decapitating the module was a collection of electrode
elements: (1) nickel metal sheeting, (2) metal hydride electrode sponge, and
(3) plastic fiber separator. It is melting of the plastic
fiber separator that leads to a permanent shorted cell:
Battery Electrode Activation
One critical aspect is to find out if water replacement can restore
It does but only after the electrodes are reactivated by a series
of charge-discharge cycles:
We also learned that heating the battery accelerates electrode
activation. But the most important lesson is we don't have to do
a full charge-discharge cycle.
A partial charge-discharge seems to be just as effective allowing
many more cycles and less power to be used.
Don't Use Manual Charger
One of our Prius Technical Group members put a charger on
his NHW20 pack without an automatic charge shutoff system:
I have found using an ordinary lead acid battery charger, even with
current limiting resistors, can do the same.
At a minimum, your battery charger should have a dV detection as well as
a temperature probe backup.
No, you can not promise to "be good," either use a automated shutoff
or destroy some batteries and then use an automated shutoff.
In Situ Testing
The Graham scanner is able to report the traction battery current and voltages
of the lowest and highest pair of modules.
By using the trend function over a set of data, the slope of the line
should be proportional to the internal resistance of the weakest
and strongest module pairs.
In this case, the slope of the trend lines are nearly identical suggesting
the differences are not too far off.
However, we don't have new cells to compare this data to figure out
the relative capacity of the modules:
Forced Charge Heating
One of the earliest observations is charging NiMH batteries leads to
module heating, it is an exothermic reaction.
The Graham scanner will report the lowest and highest module pair
A paper titled "Thermal Behavior of Small Nickel/Metal Hydride Battery
during Rapid Charge and Discharge Cycles"
Takuto Araki, Masato Nakayama, Kenichi Fukuda, and Kazuo Onda"
has figure 10 charts showing temperature response of an AA sized battery
under charge and discharge.
This excellent paper includes a model including the chemical reaction
heat functions and matches these graphs.
They had excellent agreement between theory and measurement:
These charts show 2C charge going from 60%, nominal Prius battery SOC, to
80% forced charge limit versus discharge from 20% to 40% discharge,
the ratio is:
- 6.0C increase on charge
- 4.41C increase on discharge
If we take the forced charge data from my April 28, 2008 exerperiment
and plot how many seconds it takes per 1C change, we find charging has
much shorter intervals as a function of Ahr versus the longer times of similar
The larger size of an NHW11 module compared to AA batteries means there is
some thermal lag.
The NHW11 thermistor probe is located on the top of the battery, as
about 5x further from the center of battery mass than an AA battery.
Module Voltages In Marginal Traction Packs
Two users have provided either individual module voltages or
Toyota reported voltages for each module pair.
What this shows is the end modules, those most likely to be coolest,
seem to be healthiest.
In the first case, two modules from another salvage pack had replaced
This data suggests one strategy would be to take every other end module
and swap it with every other middle module.
This would have the effect of equalizing the charges in a marginal pack
BUT this would only delay the inevitable.
Hill Descending Heating
The other primary source of charging currents is to descend
a hill at high speed using regenerative braking versus using
"B" or engine braking.
As is clearly evident, using "B" does a much better job of
avoiding charge current heating of the traction battery:
In Car Module Refurbishment
It had never occurred to me that modules could be removed from the
traction battery assembly in the car:
In this case, the JB Weld seals were a little inconsistent.
Properly mixed, JB Weld should be the same color as the
battery module plastic: