Progress Update

One of the first commercially viable automobiles, the Ford Model T, produced a whopping 20 horsepower peak from a 2.9L displacement engine for a maximum speed of 40 miles per hour. This doesn’t seem too impressive when compared to the modern engines, even those found in the most affordable of cars. So, how did we get where we are today? How did we manage to make small, low cost, low displacement engines that can provide upwards of 120 horsepower, while burning less fuel?  While the science behind the internal combustion engine has remained largely unchanged since the days of the Model T, today’s auto manufactures are equipped with high-tech tools that help them optimize the design and construction of modern engines to maximize efficiency.                                       

How did this happen? How does one increase the efficiency of an engine? There are a variety of ways to approach this problem, but one of the main methods employed in the industry involves considering how air flows through an engine. While this may not seem like an important aspect at first blush, airflow optimization is one of the principal reasons modern engines have such a high power output. Since fundamentally an internal combustion engine can be idealized as an air pump it follows that increasing airflow through the engine should yield an increase in power output. Fuel injected internal combustion engines inject fuel into air and ignite it. As a result, the maximum power of an engine is directly proportional to the amount of air that the engine can aspirate during the intake stroke and reject during the exhaust stroke. Being able to monitor airflow through an engine, particularly through the cylinder head and intake-exhaust system, while experimenting with porting and other relevant parameters is critically important in the quest for maximum efficiency and power output of an engine. The shape a port found on a cylinder head of an engine, or the inside of any pipe or tubing, affects the speed of the fluid through it. The viscous forces caused by pipe passages that bend or change shape give rise to frictional losses. In turn, energy is dissipated and the fluid slows down.                   in a new tab)

Professionally, flow benches are employed to monitor internal airflow in these sorts of cases. A flow bench is essentially a wind tunnel, but rather than measuring flow around a test piece, it measures flow through a test piece. So, wielding a flow bench, we can measure air flowing through a cylinder head and use the data such derived to optimize the engine’s efficiency. Sounds easy enough, right? There’s only one problem: attached to a professional flow bench is a price tag upwards of $20,000. This makes accessing a flow bench infeasible for amateurs and researchers on a budget.

There is a demonstrated need for flow benches in research, as evidenced by the plethora of papers whose authors use them: Ismail and Abu Bakar. (2008)Samuri et al. (2015)Siwatek et al. (2017) and Oh and Lee (2015), to name a few. There is a abundance of academic research utilizing flow benches for both traditional cylinder head testing as well as using modified flow benches to explore intake manifolds, exhaust systems and more.

In this project, we aim to create a more accessible flow bench, primarily with respect to cost. The final product should be capable of accurately measuring flow through a cylinder head. DIY flow benches do exist, but do not come close to accuracy or precision that professional flow benches provide. Our flow bench will aim to bridge that gap by providing a affordable alternative (<$300) that gives usable data for researchers and hobbyists alike.

One of the main challenges thus far has been finding ways to reduce the cost of the final product, while simultaneously maintaining a high degree of accuracy. Simply making use of the same or scaled down parts of a professional flow bench would not significantly reduce the cost, and following online tutorials and DIY designs would degrade the accuracy. However, there is an important lesson to be learned from online tutorials: not everything must be bought. Reusing parts, such as motors salvaged from vacuum cleaners to generate airflow, not only saves money, but also gives new life to parts that would otherwise end up in a landfill. And, as it turns out, vacuum cleaners are commonly discarded while their motors are still fully functional.

Deconstructing a scavenged vacuum for its motor (1).
Deconstructing a discarded vacuum for its motor (2).

Along with the aforementioned motors, one of the most significant costs associated with constructing a flow bench, is the flow rate data acquisition equipment. The flow rate is typically monitored using high accuracy pitot tubes and associated pressure sensors can cost upwards of $5000, thereby significantly increasing the cost of the complete device. Other DIY or inexpensive commercial flow benches typically employ a simple water manometer along with the aforementioned pitot tube in order to measure airflow through the test piece. However, we found a more elegant solution in another recycled part: the mass air flow sensor (MAF). Mass air flow sensors are used by modern engine management systems in order to monitor air flow into the engine, so as to precisely adjust fuel injection quantity and timing as well as ignition and cam timing to cope with whatever conditions an engine might be subjected to during its operation. Since MAFs must be accurate enough to detect minute changes in airflow for their intended application, they should also provide a viable data source in a flow bench. Further, their measurement range should also be sufficient for our application since they are capable of measuring airflow at conditions ranging from idle to full throttle at maximum RPM(revolutions per minute). Due to the prevalence of mass airflow sensors in the automotive industry, they are readily available as a used part, and can be bought secondhand for as little as $30.

Two salvaged vacuum cleaner motors.

Since the test pressure used during a flow test should match any other data sets that one wishes to make direct comparisons with, conventional analog flow benches require a human feedback loop in order to adjust motor speed such that the desired test pressure can be achieved. This increases error while decreasing usability. Further, data must be recorded manually. Professional flow benches employ a digital control system allowing for significantly increased ease of use while decreasing the time required to run a test. With this in mind, rather than using a manometer to measure the test pressure, as most low cost flow benches do, a pressure transducer (pressure sensor) will be employed. This sensor, paired with a MAF, will result in an entirely electronic data output while also minimizing cost and complexity. This will make operation of the flow bench easier for the end user since they will be presented with a single, unified interface rather than multiple simultaneous data sources. By employing only electronic sensors, our device can implement a digital feedback loop as well as coalescing data into a consolidated output.

The largest component of this control loop is the motor controller, which must take an input from the Arduino and regulate the speed of the blower motors in order to arrive at the desired test pressure. Vacuum cleaners employ so-called universal motors which are capable of running on either AC or DC. Since controlling speed using DC requires a less complex circuit, we elected to go that route. In order to accomplish this objective, we designed and built a prototype, a schematic of which is shown below:

Preliminary Schematic of Our Motor Controller Design

An initial prototype was built using entirely salvaged components and tested at low power levels in order to ensure viability of the design. Building on the success of that endeavor, we are currently in the testing phases of the final, large-scale motor controller that will be featured in the actual device. To cope with the higher power, we elected to purchase certain components new rather than salvaging all of them. However, we have encountered a problem with the larger controller: one of the two MOSFETs (a type of transistor) was damaged during a test run at full power (likely due to noise present in the input signal). We are still troubleshooting this issue in order to ensure that no further MOSFETs have their lives cruelly cut short. Moving forward, the input of the feedback loop must be considered. This loop will be implemented using software on the Arduino. In more technical detail, the basic schema for this task involves measuring the pressure using the pressure transducer and varying the pulse width (a property of a digital signal) of the signal sent to the motors until the desired test pressure is achieved. Once this occurs, the Arduino can measure the output of the MAF in order to determine the flow rate through the test piece.

      First Prototype at low power levels. Testing the large-scale motor controller.

The other major component we have been working on is the enclosure in which the blower assemblies will reside. We first considered this to be a straightforward, mundane task, but the power and torque output of vacuum cleaner motors makes mounting them securely a bit challenging. A significant moment (rotational force) is induced by the revolution of the rotor within the blower assembly. This was discovered when we tested each of the motors without any sort of housing. Without a system to restrain the motors, they tended to try to escape our testing facilities in a rather violent manner. Further, the enclosure must allow for sufficient intake and exhaust airflow for testing any item we wish, while not allowing the high pressure air to escape (if testing exhaust), or allowing atmospheric air in (if testing intake). With this in mind, we will be using a plastic storage bin along with a set of U shaped metal clamps to restrain the motors inside. The original plan was to laser cut acrylic or wood to make the sides of a box, but the low cost and high strength of premade bins makes them a more attractive option, especially after discovering the magnitude of the forces involved. Further, by procuring a preexisting bin, we reduce the number of sealing edges from twelve to four, decreasing the likelihood of air leaks.

Additional electronics work
Diagram of our flow bench design.

Works Cited

Benson, Tom. “Pilot Tube.” NASA, NASA,

Brei, Terry M. “How Low Pressure Transducers Work.” Sure Controls, Sure Controls Inc., 23 July 2013,

“Camshaft.” Wikipedia, Wikimedia Foundation, 20 Apr. 2019,

“Head Loss.” S.B.A Invent, S.B.A Invent,

Hymel, Shawn. “Alternating Current (AC) vs. Direct Current (DC).” Alternating Current (AC) vs. Direct Current (DC), SparkFun Electronics,

“Introduction.” Arduino, Arduino,

Ismail, Abdul Rahim, et al. “An Investigation of Valve Lift Effect on Air Flow and Coefficient of Discharge of Four Stroke Engines Based on Experiment.” American Journal of Applied Sciences, vol. 5, no. 8, 2008, pp. 963–971., doi:10.3844/ajassp.2008.963.971.

Lacoma, Tyler. “How Do Manometers Work?” Sciencing, 2 Mar. 2019,

Lacoma, Tyler. “How Do Manometers Work?” Sciencing, 24 Apr. 2017,

“MOSFET.” Wikipedia, Wikimedia Foundation, 9 Apr. 2019,

Oh, Daesan, and Choong Hoon Lee. “Flow Coefficient Measurements for an Engine Cylinder Head under Transient Flow Conditions with Continuous Valve Lift Change.” International Journal of Engineering and Technology, vol. 7, no. 3, 2015, pp. 1135–1142.

Samuri, M F, et al. “INTAKE PORT FLOW STUDY ON CYLINDER HEAD USING FLOWBENCH.” Journal of Transport System Engineering, vol. 2, no. 2, 26 Sept. 2015, pp. 29–35.

Swiatek, Piotr, et al. “Flow Bench Testing of Prototype Intake Manifolds for Ultralight Aircraft Engine.” MATEC Web of Conferences, vol. 118, July 2017, p. 00008., doi:10.1051/matecconf/201711800008.

Unknown, director. TenorTenor, Tenor, 4 Apr. 2017,