How to Interpret Claims of a 24% Improvement in Construction Machinery Fuel Efficiency with Nanobubbles
Recently, an interesting press release was issued by Kumagai Gumi.
It reports on a demonstration experiment where fuel consumption was reduced by mixing "ultra-high-density nanobubbles" into the fuel of construction machinery.
For us at CWM, who have built our business around nanobubble water treatment technology, this is a topic we must confront head-on.
To conclude, there are certainly parts of this announcement that make one technically skeptical.
At the same time, however, we are not taking this as an opportunity to "bash" it, but rather as an opportunity to apply it to our own verification and commercialization efforts. I will write about this sequentially below.
What Kumagai Gumi Announced
First, let me clarify the facts.
On November 19, 2025, Kumagai Gumi announced that it had conducted a demonstration test at the Otaehata Dam embankment restoration project site in Kumamoto Prefecture, with the cooperation of Anzai Kantetsu, Tokyo Systems, and Marubeni Energy.
They installed an "ultra-high-density nanobubble generator" on three types of machines—a 1.4m³ backhoe, a 0.8m³ backhoe, and a 45kVA generator—and compared fuel consumption, confirming a reduction effect in all models. Media reports state that fuel efficiency improved by approximately 24%.
The technical selling point is in the density of the bubbles.
While conventional devices were limited to 100 million bubbles per mL, this device is said to have achieved an ultra-high density of over 1 billion bubbles per mL. The mechanism is explained as follows: "by dissolving ultra-high-density nanobubbles into the fuel, the viscosity of the fuel is reduced, improving combustion efficiency."
Their focus on an inexpensive and simple device is a sound perspective.
Attempts to improve combustion using nanobubbles have been made in some fields for some time, and the recognition that device complexity and cost have been barriers to widespread adoption is also correct.
The problem lies in the numerical values and explanations beyond that.
Doubts About the 24% Figure
To be honest, I believe it is difficult, from the perspectives of chemistry and thermodynamics, to accept this 24% figure at face value as the result of pure combustion improvement.
I will divide the reasons into three points.
Point 1: There is almost no room for improvement in combustion efficiency
Modern diesel engines already burn very well. The rate at which fuel inside the cylinder is completely burned, or so-called combustion efficiency, has reached 98–99%.
In other words, there is originally only 1–2% of "unburned fuel."
The main reasons engines lose fuel energy are not due to unburned fuel, but rather heat discarded externally as exhaust gas and cooling water. Of the chemical energy of the fuel input, only around 40% is actually converted into work (output), and the majority of the remainder is discarded as heat. This distribution is determined by the theoretical limit of thermodynamics (Carnot efficiency) and cannot be moved by additives to the fuel.
Therefore, even if nanobubbles completely burned the remaining 1–2% of unburned fuel, the additional energy that could be extracted is at most 1–2%.
The 24% figure exceeds this upper limit by more than an order of magnitude. It is thermodynamically impossible to achieve this figure solely through pure combustion improvement.
Point 2: The amount of oxygen brought in from the fuel side is negligible
If the logic is "the bubbles contain air (oxygen), which helped combustion," it runs into the wall of stoichiometry. The absolute amount of gas that can be held as bubbles in liquid fuel is extremely small.
On the other hand, diesel engines originally use turbochargers to push large amounts of compressed air into the cylinder, burning the fuel in a state of oxygen excess (lean combustion).
Given this massive volume of air, it is quantitatively impossible for the tiny amount of oxygen in the bubbles brought in from the fuel side to dramatically change the entire combustion reaction. It is like trying to change the water level of the ocean with a cup of water.
I should fairly add that Kumagai Gumi's explanation cites "viscosity reduction" rather than "oxygen supply" as the mechanism.
However, even if the fuel's viscosity is reduced and spray atomization is improved, the effect it brings is limited to the range discussed next.
Point 3: There is also a ceiling to the effect of atomization improvement
The phenomenon where bubbles in fuel expand and burst at high temperatures, atomizing the fuel more finely (micro-explosion), is a real physical phenomenon. This makes it easier for fuel and oxygen to mix, contributing to a reduction in black smoke and an improvement in combustion speed. I do not deny this.
However, this is not magic either. Research into the thermodynamics of emulsion fuels mixed with water or air has decades of accumulated data, and the improvement range of thermal efficiency due to promoted atomization confirmed there is generally considered to be limited to 3–5%. 24% is far outside this range.
So, where did the 24% come from?
Based on the above, it is scientifically straightforward to think that the reality of the 24% is not "extracting more energy from the chemical energy of the fuel," but rather a combination of other factors. There are two strong candidates.
One is the cleaning effect.
For years, carbon and sludge have accumulated on the fuel injection nozzles of construction machinery, deteriorating fuel efficiency due to poor atomization.
Because nanobubbles have peeling and cleaning actions, there is a possibility that this clogging was removed, and the "original fuel efficiency" returned, appearing as a dramatic improvement figure.
This is not a story of improving combustion, but rather the minus returning to zero. The effect is real, but it is a one-time recovery, different in meaning from sustainable, fundamental fuel efficiency improvement.
The other is a problem with the measurement conditions themselves, which is the most significant one.
Doubts about the experimental method: The difficulty of measuring fuel efficiency with a backhoe
Fuel consumption of a backhoe (hydraulic excavator) is easily and significantly swayed by the hardness of the soil being excavated.
If the ground is hard or contains gravel, the load on the bucket increases, the pressure of the hydraulic pump rises, and the engine stays at high load. The opposite is true for soft, sandy soil.
Even with the same "1 hour of digging," the actual work done by the engine can vary by the order of tens of percent. Furthermore, if the soil layer changes due to excavation depth, the load also changes, and the operator's habits (pointless rotation, idling ratio, accelerator work) are added on top of that.
This is the decisive factor. The variation due to external disturbances (soil type, operation) is larger by an order of magnitude than the effect one wants to find (a few percent).
The composition is attempting to extract a signal of a few percent from noise fluctuating in a range of tens of percent. In a simple comparison of "last week was normal fuel, this week was nanobubble fuel, and fuel efficiency dropped," it is fundamentally impossible to separate whether the difference was due to the bubble effect or just because the soil happened to be softer this week.
Moreover, this demonstration was conducted in an actual field, the dam embankment restoration project.
In other words, it is the environment with the greatest external disturbances.
Unless it is shown how load conditions were aligned, how control groups were established, and how soil type variations were compensated for, the "fuel reduction" of the backhoe cannot be distinguished from the luck of the soil.
The situation is different for the 45kVA generator among the three models.
Because generators can keep rotation speed and load relatively constant, the data reliability is higher than that of the backhoe.
This is a point that should be evaluated fairly.
However, if the figure is still too high to be explained by pure combustion improvement, then suspecting the contribution of cleaning effects or measurement conditions is the logical path.
In any case, it is worth remembering that the generator's data is a much more reasonable platform.
Even So, This Technology Has Sufficient Value
I have written critically up to this point. But the essence of what we want to say is from here on.
Even if 24% is dubious, it does not mean this technology has no value. In fact, it is the opposite.
Even if the true effect is conservatively estimated to be a few percent, it holds enormous value.
The reason lies in simple multiplication.
Fuel costs are a dominant term in business costs for industries such as shipping, mining, large-scale infrastructure, and stationary power generation. A reduction of a few percent in fuel here is a direct impact on profit, i.e., a cost reduction of a few percent.
For example, for a business that consumes hundreds of billions of yen in fuel annually, a 5% reduction is on the order of over a billion yen annually. In industries with operating profit margins of a few percent, this has the same effect as significantly increasing sales. Moreover, a reduction in fuel consumption leads directly to a reduction in CO2 emissions and a contribution to carbon neutrality. A few percent is not "small." It is more than enough to drive business.
That is why our stance becomes this: Do not hype "24%!", but prove a reliable and sustainable few percent using indisputable methods.
Flashy numbers are met with caution by experts and are a double-edged sword that damages trust.
Modest but robust numbers are far stronger when dealing with large industries.
Verification Design CWM Must Undertake
Then, how should we prove it? The weaknesses in Kumagai Gumi's demonstration are the reverse of our design guidelines.
First, the initial verification will not be built on field construction machinery, but on stationary generators.
This is because the load can be fixed electrically.
If a resistive load bank is connected to a diesel generator and multiple load points such as 25%, 50%, and 75% are fixed, soil quality and operator habits do not exist.
This becomes a platform where a few percent signal can be targeted with a few percent measurement accuracy. While noise was larger than the signal by an order of magnitude with the backhoe, this is reversed here.
Second, we will grasp not only "that fuel efficiency dropped" but also "why it dropped" at the same time.
If it is stationary, fuel consumption can be measured with high precision using the weight method (placing the whole tank on a scale and measuring the decrease in mass per time), and exhaust gas composition, black smoke (smoke meter), and exhaust temperature can be measured in parallel. We can explain the mechanism with data: Did combustion become uniform? Did unburned components decrease? We will move to the side of disclosing parts that Kumagai Gumi is hiding.
Third, we will include the energy used to make the bubbles in the balance.
We will measure the power consumption (kWh) of the nanobubble generation device and output the net balance after subtracting it from the fuel efficiency improvement. If this is left ambiguous, one falls into the trap of "apparent fuel efficiency improvement."
CWM's Nano-Inset makes energy efficiency (bubble generation amount per kWh) its strength. We can show that advantage in figures specifically for fuel applications.
Fourth, we will separate the cleaning effect and combustion improvement.
We will stabilize the device with sufficient break-in operation before the test, and measure the temporary recovery due to cleaning and the sustainable improvement in the steady state separately. The former has value as the former, but mixing the two and calling it "fuel efficiency improvement" is scientifically dishonest, and we will not do that.
With this design, even if the effect is zero, the value of "having shown with high precision that it is zero" remains. With field construction machinery, it tends to lead to the worst conclusion of "we don't know if there is an effect or not," which is a decisive difference.
Closing
Kumagai Gumi's announcement has issues with how the figures were presented and the experimental environment. That is something that should be pointed out frankly as an engineer. But what this incident shows is rather the market movement itself where major general contractors have begun to work in earnest on the theme of combustion improvement using nanobubbles.
We at CWM intend to take this story as a fortuitous opportunity. We will not chase flashy numbers, but build up robust verifications that can explain the mechanism, disclose the balance, and be replicated by third parties.
When we can prove a few percent of improvement that no one can deny, it will undoubtedly be a major contribution to huge industries such as shipping, mining, and infrastructure. Competing not with flashiness, but with certainty. That is the path CWM should take.
