Verification and validation testing of aerospace components, notably pumps, valves, engines, is essential to ensure safe flight. Adequate testing requires exact replication of actual geometrical representation of component position within an aircraft. Further, it is quintessential to simulate environmental conditions surrounding aircraft during the flight. If the component is not subjected under real application conditions during testing then it is never justified if it is suitable for use in service. Testing of aerospace components should be done both at qualification and productions stages. For qualification, components are run to the extreme conditions. It is often necessary to run components above the actual application conditions. Sometimes, components are run to excessive conditions until the components are destroyed. The maximum limits of the conditions are then recorded and used as a benchmark to determine the worst conditions for the flight. For production stage testing, the conditions are only few notches above the normal operational conditions. During production testing, components are batch tested, stamped and released for installation into the aircrafts.
The testing of valves and pumps have always posed a challenge to the manufacturing engineers. These components drive hydraulic fuel through the engine. These components are run at high pressure and flow during the flight. The work in this report demonstrates my recent approach to achieve the testing of aircraft components in a safe and a cost effective manner.
This article demonstrates some methods to 1) design a suitable test rig solution for aircraft valves/pump testing. 2) to produce desired flow and pressures and develop means to measure these parameters 3) to develop a suitable strategy for simultaneous closed loop control for all fuel supply lines i.e. supply/return.
Aerospace servo valves usually have supply and return port. In addition, these valves have couple of control ports which are connected with circuits where controlled flow and pressure is required. It is outside the scope of this article to target a particular model valve. The aim is to present a test solution model for such aerospace valves. These hydraulic ports could be supplied with fuel at full range under controlled conditions. Pressure and fuel flow sensors could be added in to the test rig circuits to collect measurements. Those measurements could then be compared with models to determine if aerospace component under test is suitable for use in service.
There are various mechanical and electrical components that are required to achieve a full test solution. The main components are described here:
Internal Gear Pump
A suitable internal gear pump was used to pump fuel into the hydraulic circuit. Internal gear pump provided smooth output flow which was pulse-free and proportional to the speed of the drive motor. Within internal gear pump, Drive motor is connected to internal gear which pushes the external gear as in the animation above.
For a machine like test rig solution, accumulator was strategically placed after the pump to further absorb any pulsating flow and pressure out of the pump.
Hydraulic fluid pulsations were further reduced by another component that was placed along with accumulator above.
Once out of pump and through attenuator, series of valves, pressure sensors, flow sensors were used
Before entering main hydraulic piping, a good quality filter was installed for fuel cleanliness.
Usually these are normally closed and are pressure-limiting valves. When pressure at input reaches the valve setting, the valve starts to open to release the pressure by letting flow through the return line back to the tank, throttling flow to limit the pressure rise
These are also called check valves or one-way valve. These valves allowed fluid flow only in one direction. The fluid flow is prohibited in the direction of the arrow in the symbol below and only flows in the opposite direction.
Manual Ball Valve
Manual ball valves are simply manual valves that stop or start flow. If valve position is at 90 degree to flow pipe then flow is stopped. On the other hand, if valve handle is aligned along (parallel to) the pipe flow then the flow is allowed.
Pneumatically actuated ball valve
As opposed to manual ball valve, pneumatically actuated ball valve were ideal to achieve automated control via digital output. A software button was used to switch digital output which then opened/closed the valves automatically without requiring manual intervention.
PT100 is a temperature sensor and was used to measure the fuel temperature
Pressure Sensors were used to measure pressure in bars or psig. These were also used as closed loop control feedback sensor to regular pressure in hydraulic supply circuit.
Proporional Relief Valve
These were essential to achieve required pressure dynamically. Depending on the electrical current output ( Analogue Output), the orifice of these valves was continuously adjusted to achieve required pressure.
In addition to the above components, a hydraulic machine could also be fixed with a heat exchanger. A heat exchanger routed the hydraulic oil through narrow passages with large surface areas that transferred heat from the hydraulic oil to fluid (water for this design) that carried the heat away
A test rig was designed to incorporate the above hydraulic components to achieve precise and regulated control on the supply line, return line as well as fuel flow through the internal gear pump. As a matter of fact, three closed loop control sub-systems were required and due to the interconnected piping network, a situation was developed where the three control loops were found conflicting against each other. Such unstable behaviour was expected until test rig closed loops were fully tuned.
There are identified ways to achieve reliable closed loop control. For instance, often FPGA is used as a controller to run three closed loop control independently rather than sequentially which is expected when run on a micro-controller or programmable logic controller PLC. Usually, code written in text-based programming language would run in sequence starting from the first instruction. There are ways to program and code to achieve parallel processing but it was concluded that those established standards would never read as independent, parallel software module unless written in a graphical programming language.
To achieve the technical challenges offered in the above scenario where multiple closed loop controls were required to run independently, it was identified that one of the most cost-effective approach was to write control software within labview. Labview was found to be based on data-flow paradigm and vis (subroutines) could be placed on the block diagram (IDE coding screen) in parallel fashion. Labview automatically assigned available processor thread to run control loops independent of each other. Here is the IDE view of multiple loop running in parallel fashion.
PID CONTROL SYSTEM
PID control was found to be the most common control algorithm in industrial control applications. The techniques presented here produced zero overshoot and smooth ramping of supply/return pressures as well as fuel flow through pump. Hydraulics supply and return pressure was controlled via proportional control valves from Wandfluh and Masoneilan. Pressure transducers were obtained from Durck. NI PXI controller was used for rig control, data acquisition, logging and PID control. Software was written in labview.
Non-Linear Control System and Calibration
It was found that control valves did not have linear behaviour. Accordingly control output from PID would result in excessive opening/closing of valve orifice that resulted in unexpected fuel pressure in the supply and return lines. In order to mitigate the effects of non-linear valve profile, PID Auto-tuning and ten-point calibration of control valve was performed.
Calibration involved applying known pressure in the hydraulic system and recording the values of pressure on the output of control valves. The raw sensor value as well as calibrated scaled PSI values were then logged in a calibration table. Finally, the calibration was applied on the demand set-point and pressure transducer feedback. By applying calibration based on control valve profile, PID generated only the control output that was required to minimise the mismatch error between demand set-point and pressure transducer feedback.
It was discovered through experimentation that one of the most effective way of applying calibration was by inserting instantaneous raw pressure transducer value into calibration table and interpolating the table to find corresponding calibrated engineering value.
PID control required demand set point value which in the above case was the desired supply or return pressure. PID control valve also needed pressure transducer feedback.
It was found from experimentation that there would be noise on the amplified feedback signal which in the above case was pressure sensors. The entire test rig started to oscillate vigorously which was expected as a fast PID control system would immediately respond to the change in feedback signal. It was concluded that the most effective approach to prevent test rig from oscillation was to round the feedback signal to two decimal places. In addition, a simple filter was applied on the feedback signal to eliminate mains 50-60 Hz noise.
Here is a block diagram showing a PID control that was developed to achieve zero overshoot
PROPORTIONAL, INTEGRAL AND DERIVATIVE PID VALUES
Each PID control loop was run under PID autotune mode which produced starting PID values. After which, further tuning was performed by adjusting P, I and D values. It was discovered through experimentation that the fastest approach to achieve the best PID values was by starting with I and D value of 0 and slowing increasing P value from 0.
Proportional P value determined how fast supply pressure would reach demand value but as P value was slowly increased, the rig started to eventually oscillate. P value was then gradually reduced until minimum oscillation was produced.
After tuning P value, Integral I value was tuned to eliminate the oscillation and reach stable supply pressure.
COMPLETE TEST RIG CONTROL WITH MULTIPLE PID LOOPS
Each control loop was tuned at required rig flow. Once tuned, control loop dynamically adjusted controller output to achieve required demand pressures on the supply and return line. The user interface provided an excellent training academy to learn the closed loop control. Here is the snapshot.
If you need assistance with designing a controller for your hydraulic machines, then I might be able to assist. I write regularly on my experience and engineering work on www.easybodge.com. You can also subscribe to my blog there to get instant alerts.