·Home ·Table of Contents ·Aeronautics and Aerospace

Automated high integrity ultrasonic examination of composite rocket motor cases AC Whittle - Rolls-Royce, A Whittle - Peak NDT Ltd., S Bruni - SE.CO.SV.IM, A Gatta - Fiat Avio Contact

Summary

An ultrasonic system for the inspection of the rocket motor case of the Zefiro launch vehicle has been installed and commissioned at Fiat Avio's plant in Colleferro. The system is to a similar design concept as the system supplied by Rolls-Royce MatEval for inspecting Ariane V rocket motor components (reference 1) and is based around a high performance Micropulse 3 multi-channel flaw detector.

This paper reviews the inspection requirements and the ultrasonic techniques devised to satisfy them. Results from the first inspections are also presented. The work is significant in that not only are the defects of concern detected, accurate measurements of both case and insulation thickness are possible. This is important in the trade-off between weight and strength - any savings in weight allowing for an increase in payload.

Introduction

The filament wound carbon fibre construction of the Zefiro motor case, currently being pioneered by Fiat Avio, poses special problems for non-destructive testing. Thermally insulating rubber, which is highly attenuating to ultrasound, is laid up on a mandrel and vulcanised. Subsequently the case is 'wound' on the outside and then cured. The case is essentially a cylinder (approximate dimensions 2m diameter x 4m long) with dome-shaped ends (figure 1). Circular apertures in the domed-ends provide restricted access for personnel and manipulative equipment. The inspection surface is fabric-like in appearance, difficult to wet using bi-distilled water which must be used as couplant and the use of wetting agents is not permitted. Strong interweaving of the carbon fibres and the presence of excess resin on the outer surface of the domes present further problems.

Fig 1a: Schematic diagram showing manipulator deployed in raised position for inspections from the inside

Fig 1b: Schematic diagram showing manipulator deployed in lower position for inspections from the outside

Both the thermal insulation and the filament wound case must be inspected for delaminations and voids and also the integrity of the bonding between the two checked, achieving as near 100% coverage as possible. The thickness of both the case and the thermal protection must also be measured. The requirement for the case and thermal insulation thicknesses to be measured individually and the necessity to determine the location of any defect through wall, means that the through-transmission method, conventionally used for the ultrasonic inspection of filament wound components, cannot be used. Separate pulse-echo inspections must be applied both from the inside of the case and the outside.

The inspection may be required to be carried out before and/or after hydro-test. When carried out after hyrdo-test, inspection from the outside has been found to be hampered by the presence of micro-porosity in the case. This develops during hydro-test near the outer surface and is not of structural concern. It is, however, still a requirement to inspect the remainder of the case wall thickness and to measure the overall thickness. Additional ultrasonic tests from the inside had to be devised, triggering off the bond-line between the thermal protection and the carbon fibre case, to perform this function.

Ultrasonic techniques

The pulse-echo method is used for delamination detection, thickness measurement and disbond detection from both the inside of the motor case and from the outside. Zero degree beams give perpendicular incidence on flaws and the bond-line alike, allowing defect detection by specular reflection. Unlike the Ariane V inspections, where owing to the acoustic mismatch between the thermal protection and the steel case the phase of the signal rather than its amplitude had to be monitored, it is possible to use an amplitude-based technique for disbond detection. The presence of a disbond is indicated by an increase in the amplitude of the bond-line signal.

The flexibility of the Micropulse-based inspection system is exploited in that each probe is fired a number of times to:-

• monitor coupling,
• check for delaminations and
• measure thickness and detect disbonds.

In the absence of a motor case containing artificial defects, the capabilities of the proposed techniques for defect detection and their accuracy for thickness measurement, were established through test-piece trials.

Inspection from the inside
For inspections from the inside, the 0° compression-wave rubber delay-line probes, specially developed for the Ariane V inspections were used. These allow the coupling to be monitored by gating the signal from the end of the delay-line and also improve the near surface defect detection capability. However, for the Zefiro motor case inspections, the fact that the thermal insulation material has different properties to that inspected previously, means that, even when the coupling is good, the signal from the end of the delay-line does not disappear completely. There is about a 10% difference between the velocities in the two rubbers.

The use of 2MHz broad-bandwidth transducers allows inspection through a range of thermal protection thicknesses by changing the filtering of the received signal. A 1MHz filter will give superior thickness measurement accuracy at shorter range, whereas a lower frequency filter will allow greater thicknesses of rubber to be penetrated.

In order to achieve the required thickness measurement accuracy (better than 0.5mm), careful attention must be paid to calibration and to the measurement of the ultrasonic velocity and the delay-line thickness. Half-wave rectification for coupling and thickness tests eliminates the problem where successive half-cycles may be of similar amplitude - sometimes one half-cycle may be reported as the largest in the gate and sometimes the other. This can lead to an error in thickness measurement of as much as 0.5mm, depending on the probe frequency. If the coupling tests (signal from the end of the delay) are set to report in microseconds, rather than millimetres, then the reported range may be used directly for the value of the delay for all other tests with that probe. This has been found to simplify significantly the process of calibration, which is particularly important where an array containing a large numbers of similar probes is used, and produces superior thickness measurement results.

Inspection from the outside
Similar techniques are applied from the outside except that 5MHz immersion probes are used. The amplitude of the interface echo (water path / carbon fibre) may be used as a coupling monitor. Again the delays for the thickness and delamination tests may be set from the range of the interface echo, if it is reported in microseconds. Note, inspection from the outside is only possible over the cylindrical part of the motor. Strong interweaving of the carbon fibre at the domed ends coupled with the presence of globules of resin on the surface prevented the inspection from the outside using the present system.

Test sensitivity and the use of DAC curves
In order to achieve a uniform testing sensitivity throughout the testing range, Distance Amplitude Correction (DAC) must be used. The amplitude of the back-wall echo from flat test-pieces of a range of thicknesses was used to define the DAC curves and to establish the ultrasonic test sensitivity. Note that for DAC curves for inspecting through the thermal protection more than one curve was required to cover the entire testing range as the thermal protection material is highly attenuating.

For inspection from the inside of the motor case the sensitivity was set from the bond-line signal. With the required DAC curve selected the bond-line signal was adjusted to 80% full screen height for both 'delamination' and 'thickness' tests. There was no need to add any 'up-gain' for 'delamination' tests as a delamination would be expected to behave as a free surface, giving a larger signal than the bond-line signal. This sensitivity level was confirmed as being appropriate in measurements on representative test-pieces.

The inspection sensitivity for the outside cylinder inspection is set from the back-wall echo of the skirt and an appropriate 'up-gain' is then added.

Detection of defects in the case from the inside
In measurements on test-pieces containing representative defects it was shown to be possible to detect defects within the carbon fibre case through the thermal protection (figure 2). This proved to be particularly important for inspections after hydro-test where the case was found to be impenetrable by ultrasound from the outside.

Fig 2: A-scan re-construction.Detection of defects in carbon fiber case through 26mm of thermal protection.

'Echo-gate-trigger' tests, triggering off the bond-line signal (i.e. the interface between the thermal protection and the carbon fibre), were set up. An adequate amplitude of interface echo to trigger the test, regardless of thermal protection thickness, was ensured by applying the appropriate thermal protection DAC curve. By specifying the velocity as that appropriate to carbon fibre, any indications found would be reported at a range, from the interface, in millimetres of carbon fibre. In this way direct measurements of the case thickness were possible from the inside.

These tests also served to increase the coverage for the inspection of the carbon fibre of domes. From the outside inspection had not been practicable, using the contact transducers proposed, owing to the strong interweaving of the carbon fibres and the presence of the globules of resin on the inspection surface.

Inspection implementation

The concept of the Zefiro motor case inspection system is similar to that of the previous system supplied by Rolls-Royce MatEval for inspecting the rocket motor cases of the Ariane V launch vehicle (reference 1). The case is allowed to rotate continuously as a probe-pan containing an array of probes (fifteen for the cylindrical portion) is scanned along at bottom dead centre. The use of a probe array reduces the inspection time. Whereas for the Ariane V inspections, testing was required from the inside only, for the present inspection it is required to be carried out from the outside as well (figure 1).

In order to facilitate inspection control, the inspection is divided into 'Zones'. These are principally defined by the local geometry. There were originally six zones - the forward and aft domes and the cylindrical section - inside and the outside. In the event, inspection of the domed ends was not possible from the outside. The strong interweaving of the carbon fibres and the presence of globules of excess resin on the surface prohibited the application of the contact techniques proposed.

Careful consideration of the manipulator requirements allowed the inspection of the remaining four zones to be carried out with a single manipulator. The manipulator is of adjustable height. It comprises a main beam with motor and belt drive to move the inspection head. Positional information is provided by encoders. The beam is mounted on a support by means of a slide-way and a hoist raises and lowers it according to the needs of the inspection. In order to accommodate the inspection of the domed ends (from the inside) and to allow insertion of the beam, both into and underneath the motor, the vertical mast is mounted on a rotary table to allow it and the probe-pan to be rotated. The extension of the mast and the rotary movement is again motorised and encoded. All the axes are controllable from the computer. This allows control of the manipulator, when it must be fitted with a smaller probe-pan and longer mast, for following the complex curvature of the dome ends as well as for the inspection of the cylindrical sections. Limits switches on all axes prevent the manipulator from moving outside of its zone of safe operation.

Once deployed, two steel fabricated trolleys support the beam. Both are fitted with wheels, to run in the existing rails and a bolt down mechanism. The Micropulse, computer and motor control cabinet are mounted on the 'main' trolley as is the pump and vacuum system for the supply and retrieval of bi-distilled water, used as couplant. The second trolley is located at the small aperture end of the motor with two positions for beam location.

The manipulator is stored with the beam retracted and supported at its mid-point by the 'main' trolley with the mast and probe-pan removed. Once the trolleys have been craned into position on the rails the height of the beam may be adjusted, using the hoist, according to whether the inspection is to be carried out from the outside or the inside. When the beam has been locked into place (vertically), the trolley may be wheeled into position along the rails and bolted to the floor. The beam can then be driven out on its slide-way either through, or under, the motor, from the pendant control. It can then be located and locked into position on the support frame. A tie-bar prevents the rotary table from being pushed out with the beam. The appropriate mast and probe-pan are then installed.

Once the manipulator has been datumed, using the automatic datuming sequence provided in the software and the various calibrations performed, the inspection can commence under software control.

Inspection results

Test-piece trials
Test-pieces were provided by Fiat Avio for proving the techniques. These contained artificial defects of a variety of sizes and were examined manually. When inspecting from the carbon fibre side delaminations and disbonds of 8 x 8mm (the target size for defect detection) were detected at the scanning sensitivity and pitch. When inspecting from the thermal protection side, again delaminations of 8 x 8mm were be detected. Disbonds needed to be larger than the beam diameter in order to give an increase in signal amplitude. However, the complementary inspection from the outside provides the required defect detection capability for the cylindrical parts.

The thickness measurement accuracy on flat carbon fibre test-pieces ranging in thickness from 4.2mm to 18mm was generally better than 0.1mm. For the thermal protection the accuracy over the thickness range 2 to 45mm was generally better than ± 0.15mm. These both exceeded the target accuracy required.

Motor inspections
A partial inspection of one motor was carried out in April 1999 and a full inspection of another was carried out in March 2000. In the 1999 inspection a small indication was identified close to the inspection surface. After careful examination of all the plot types, this was interpreted as being from a delamination in the thermal protection. The flaw was subsequently identified by manual ultrasonics and also visually - it was surface breaking. This provided good evidence for the capability of the overall system. In the inspection carried out in March 2000, the improved calibration procedures devised for the delay-line probes used for inspections from the inside, resulted in extremely reproducible thickness measurements across all fifteen probes in the array. Additional tests were added to allow the carbon fibre case to be inspected through the thermal protection, as the inspection was being carried out post-hydro-test. These enabled the thickness of the case to be measured and the presence of any flaws within the case to be identified. Some flaws, disbondings in the carbon fibre between the skirt and the dome, were indeed detected. These had been identified previously in manual inspections from the outside prior to hydro-test, but were obscured from the outside after hydro-test by the development of micro-porosity close to the surface. Again this provides evidence of the overall capability of the system. The overall inspection times achieved were significantly shorter than those foreseen originally

Conclusion

Careful attention to ultrasonic technique and manipulator design have enabled the system for the inspection of the Zefiro rocket motor case to meet the design intent.

References

1. Duncumb AC, Bruni S, Gatta A 'Ultrasonic Inspection of Ariane V Components' Paper presented at 6^th National Conference of the AIPnD, Milanofiori, 28 - 31 October 1990.

© AIPnD , created by NDT.net

|Home|    |Top|