EV Engineering News

Can Tanktwo redefine the battery pack with big data?

A Finnish-American start-up with ambitious goals wants to change how we think about batteries.

The concept of an EV that supports the quick removal and replacement of a discharged battery pack was all the rage about 5 years ago. Since then, however, various systems have been attempted and subsequently abandoned by companies like BetterPlace, Renault, and Tesla.

At first glance, Tanktwo’s ideas seem like just another gimmicky concept for battery swapping. But once we got beneath the surface, we realized the company’s plan is much bolder. Tanktwo doesn’t just want to resurrect the battery swapping discussion – it wants you to completely throw out any preconceptions of batteries.

If you can do that, the company promises big cost savings right out of the gate compared to traditional battery pack designs. In fact, it sees the battery swapping scheme more as future added value for the consumer. “What will really make this thing work is the business model for our industrial partners,” Tanktwo CEO Bert Holtappels told Charged.

Smart cells, randomly packed

It all starts with what the company calls “string cells.” These are ellipsoid balls that contain typical lithium-ion battery materials as well as electronics for sensing, measuring, switching, routing, and communications. The outer surface is a combination of plastic and conductive materials.


Tanktwo says that the electrode and separator materials are cut by suppliers in a custom shape and wound in such a way that the housing is filled with almost 100% efficiency. By coincidence, the surface area of electrode material in the standard 42 mm-long string cell is nearly the same as that of the common 18650 cell, so the average capacity is almost identical, all other things being equal. Tanktwo’s system is agnostic to lithium-ion chemistries, and the company is working with several cell suppliers to incorporate their standard formulations into small and intelligent string cells.

Tanktwo’s system replaces a vehicle’s battery pack with a container filled with several thousand randomly arranged string cells. Through the conductive material on the surface of the enclosures, string cells form contacts with one another. The terminals of the internal electrochemical cell can be connected to any of the contacts, controlled by the internal processing unit. The system maps the randomly located string cells, and algorithms calculate the optimum connection routes to make the cells work as one logical, large battery pack.

Tanktwo QA4

Holtappels spent 15 years working for Nokia, so it’s no surprise that the system borrows many concepts from the telecom and IT industries. With a tank full of string cells distributed randomly, there are thousands of different possibilities for routing connections. Using algorithms based on internet routing protocols, the cells will automatically connect to make branches and bridges – continuously changing to optimize for whatever happens to be the most efficient. “It’s very similar to sending an email from London to San Francisco,” explained Holtappels. “Today it could go through New York, tomorrow through Miami, and then through Tokyo.”

Tanktwo QA3

Tanktwo plans to maintain historical data for every individual string cell ever produced, and the company says the practice has already led to an unprecedented amount of observed battery behavior in terms of both precision and granularity.

By combining traditional cell chemistry modeling with learnings from analyzing large amounts of individual cell data collected in the field, the company will be able to more accurately compute the residual life of each cell. Combine that knowledge with the ability to control each cell individually, and that’s where the magic happens, Holtappels says. “The modeling of how cells behave at any point of their lifecycle, in any kind of environment – that insight, and providing our customers the ability to act upon it, is our key value add. It allows our customers to make more efficient use of the technology and raw materials that are already in existence.”

On top of all that, Tanktwo also claims the ability to quickly swap the cells. The company says it uses mainstream air conveying systems and holding tanks to complete a full tank swap in about 3 minutes. Due to the unique geometry (ellipsoids with a specific semi-axis ratio) the random packing density is almost as high as theoretically possible, leaving little opportunity for cells to rattle around in the vehicle’s tank. Tanks an also be partially filled, using a bladder system to hold together less-than-full loads. The company’s grand vision is a future where EV drivers will no longer need to decide how big a battery they might ever need, because it is trivial to add capacity later. “Buying an EV with a massive battery pack and the associated sticker shock will no longer be the only cure for range anxiety,” said Holtappels.

Show me the savings

Tanktwo’s vision represents a pretty big change in the EV ecosystem, and one of its biggest challenges is telling a broad and complex story. However, Holtappels believes the simplest way to get the point across is to emphasize the upfront savings an EV manufacturer can pass on to its consumers by using intelligent string cells.

So, how does this save costs? Tanktwo says that its system can make more efficient use of all of a battery pack’s capacity. Traditional battery packs are designed with a lot of extra capacity that is not accessible while driving. For example, LG Chem Power CEO recently told Charged that when it helped GM design the first-generation Volt, the goal was about 40 miles of electric range, and at 200 Wh/mile, that’s 8 kWh. But the installed capacity is 16 kWh. This is because they allowed for 30% degradation over 10 years of life and, to protect cells from safety hazards and premature wear, the Volt only uses about 70% of that capacity window. If you put those two factors together, you get about 50%, so you start with 16 kWh of nominal capacity.

Tanktwo claims it can reduce that extra capacity significantly. Holtappels explained how:

This is possible with string cells because of several effects that amplify one another. First of all, in a traditional pack, module or cell variance is the enemy, to be avoided at all costs. Tanktwo string batteries, however, work efficiently even if cell performance varies widely. When higher variance can be tolerated while retaining most of the efficiency, the state-of-charge (SOC) limits can be widened.

Not only can the spread be wider, but a significantly deteriorated cell, like one that has lost 70% of its design capacity, can still contribute to pack capacity in a meaningful way. This means that string cells, even when employing identical battery chemistry, have a longer useful life or can be pushed harder, hence a pack can be smaller. For example, a cell with elevated internal resistance will contribute disproportionately in low output power mode (while cruising), and the ones with lower internal resistance allow more current draw when peak power mode is active (during acceleration).

Tanktwo QA5

This means that, in our system, the pack capacity is always the sum of the cell capacities. This is not the case with traditional packs, which is why they need to employ techniques for cell balancing. When you discharge a regular pack and one cell or module reaches maximum depth-of-discharge, your pack is considered flat even if the other unbalanced cells still hold charge. We discharge each individual cell to its specified DOD.

Also, as a regular pack ages, it needs more balancing, which is simply shunting (burning off) charge in the stronger cells. But ours does not need balancing – on the contrary, we use the spread to our advantage.

Tanktwo QA6

So every string cell contributes to the best of its ability at all times. Old and new string cells, cells of different capacities – and even chemistries, in theory – can be mixed without limitations. As the pack ages, this increases in importance because the difference between the best and the worst cells grows due to normal variation. Our system tries to organize the strings such that the power output over a charging cycle is optimized. So when older cells are not asked to deliver the same amount of power during acceleration, this gives them a longer useful life.

Using wider SOC windows can accelerate wear, but it becomes a business case in which an increased failure rate can sometimes be profitable. And we make capacity restoration very easy. So instead of over-dimensioning to prevent warranty liability, you can restore capacity in a few minutes only if and when it happens. We make a shop-vac-like analyzer that will remove string cells that do not meet certain criteria. Replacing the lowest performers with new cells happens in minutes, restores pack performance to the required level, and costs a fraction of a total pack.

Tanktwo QA7

We can also get excellent performance just using air cooling. String cells can self-throttle, so heat dissipation can be equalized by giving cells a lower duty cycle when located in any hot spots of the pack. Instead of dimensioning a cooling system such that even the smallest nooks and crannies cannot overheat, our cooling is dimensioned proportionally and weighs much less. Those few cells that have less than perfect cooling are not overheating, because they hold back when the target temperature is reached. Liquid cooling is a possible option for high-performance niche applications. We’ve cleared a path towards using hermetically sealed string cells and a non-conductive coolant, but it is not our current priority.

Overall, we significantly save weight, which compounds all these effects. And, finally, all this is sensors, software and communication – it produces data that can be acted upon. Everything can be changed and improved over the air as the cell component suppliers are able to characterize things better, as more historical data becomes available, and as new algorithms and methods are developed.

Assuming all this works, wouldn’t adding intelligence and switching capabilities to every cell increase the pack overhead and cost? Holtappels explained that the potential savings outweigh additional costs.

The numbers work out in our favor for a few reasons. The components added to make cells intelligent cost only cents per cell in volume, as the CPU and memory footprints are very modest by today’s smartphone standards. But the immediate efficiency gains I described are on the order of tens of percent. So from day one our customers win.

Tanktwo QA8

Then there are other financial benefits and possibilities beyond the upfront cost. For example, Detroit teaches us that one of the most expensive parts of the car business it that you have to keep certain financial provisions for covering future warranty costs and liabilities. This is what leads them to over-dimensioning the battery in the first place. The last thing they want is a lot of EV owners asking for a new battery under warranty after 6 or 7 years.

That problem will go away if you have a battery that is made with our technology. You can literally choose to push the battery and knowingly drive 20% of cells into the ground over a five-year period. Because the cell swapping is so easy, dealerships can have a device for maintenance purposes that separates the good cells from the bad cells. Let’s say you have 80% good cells and 20% bad. You just replace the 20% bad cells with new ones. So, instead of over-dimensioning the battery at the moment of manufacture, you can deal with any sort of liabilities for guaranteeing pack performance by repairing the pack in a 5-minute service cycle, if needed. It’s a totally different kind of thinking.

Tanktwo QA9

Most of the depreciation in today’s EVs comes from the battery pack – no one challenges it. It’s a big investment in raw materials for batteries to store energy, so why would you want any to go to waste? In hindsight, I think people are going to say that it was really obvious that we needed a more intelligent system to connect the dots. The time is right to make batteries smart.

A swappable future

All these factors combine into what Tanktwo believes is a great opportunity to succeed where others have failed at building a battery swapping ecosystem. The company thinks it can circumvent the chicken-and-egg problem by offering a high-performance battery solution that has lower costs from its first day on the mass market. Without a single quick swap station, a string battery-equipped EV can function like a normal plug-in and charge from the electrical grid. Holtappels explained that the intelligent cell technology also solves another critical problem of battery swapping scenarios.

I think one of the main reasons why Better Place failed was because it forgot to think through the reverse logistics. The battery pack is worth several thousand dollars, or even tens of thousands. No one wants to end up with a used battery pack after a swap if you can’t accurately quantify the residual life of that energy carrier. When you do a cell swap you want to make sure you get a new set of cells that is at least as good as the one you gave in. Now that security is there, intelligence is there, and there is the data to back it up, a system can be built that keeps everyone informed and honest.

Because our technology makes it possible to predict the expected performance for each cell, our customers can now take a position on the monetary value. They can make sure that the cells which you’re going to get in a swap are equivalent and of the same value as the cells that you give in. There might be tens of dollars of difference for the whole pack, but you’re going to pay the difference or get a refund. A provider now might even try to upsell EV drivers more capacity. So the removal of the barrier to trade is actually the mission of our business.

Building a coalition

The big auto OEMs make decisions very slowly, and plan roadmaps that stretch out a decade, and while Tanktwo says it has got their attention already, it’s a long way from getting cars on the road. The company initially built small-scale prototypes to demonstrate the hardware and software. Recently, it’s upped the ante and struck a deal with Valmet Automotive, former manufacturer of the Fisker Karma. The same engineers who built the Karma are now building a functional showcase vehicle powered by 3,500 string cells. Holtappels would not say which vehicle is being used, but he says it is not a Karma. In addition to passenger vehicles, the company will target fleet, utility and industrial applications.

Tanktwo is also assembling a consortium of industry stakeholders who believe in a future for smart or distributed batteries in any form, with intelligence or information-carrying capability. Anyone who thinks this is a growing product category is welcome to join and discuss topics like the information security and safety of the systems.

This may all seem like a bit of a long shot, but some of the biggest champions of the EV industry have a long history of challenging the status quo. It’s healthy to encourage people to think about why we’re designing systems the way we are. Is there a better way to do this if we start with a blank sheet of paper?

If you study the history of disruptive ideas, you realize that they never come from within the industry that’s about to be overturned. Instead, they are often concepts borrowed from one industry and applied to another. Like, perhaps, borrowing algorithms from the IT and telecom industry and using them for distributed battery management.


This article originally appeared in Charged Issue 21 – September/October 2015. Subscribe now.

Create Account. Already Registered? Log In

Virtual Conference on EV Engineering: Free to Attend

Don't miss our next Virtual Conference on April 15-18, 2023. Register for the free webinar sessions below and reserve your spot to watch them live or on-demand.


EV Engineering Webinars & Whitepapers

EV Tech Explained