Help a student with aircraft SHM?

[ibushong]

Hello all,

I am an intern in charge of prototyping a wireless sensor network to be used to monitor the structural health of aircraft, specifically airtankers used in firefighting, like the C-130 and P-3 (focusing on wing fatigue). The system would be used in flight, either logging the stress incurred for analysis and inspection on the ground, or providing active feedback to the pilot. Now, since I have no experience with aircraft mechanics, my duties basically only concern the design of the network and relevant hardware/software. However, I wanted to ask here to get more info about what data/sensors would be best, so that I can get a better idea of what I need to aim for (and if it's even plausible!).

I am guessing that a network of strain gauges along key parts of the wings would be beneficial. Accelerometers and vibration sensors might also be useful? I have read several papers on sensor networks in civil structures, but when it comes to measuring wing fatigue, what are the key indicators? Strain amount? Number of cycles? changes in frequency response? modal shifts? Would sensors confined to the center wing box area be enough? (I need to know how far the network will extend).

As I said, I am only the first step of this process, so I'm just trying to collect as much information as I can to help it along. So if you all have anything you think would help me out, I'd love to hear it! Thanks!

 

[ibushong]

(Don't see a way to edit posts...) Here is an overview of a failure that the system should be able to help detect:

http://www.iasa.com.au/folders/Publications/Legal_Issues/PearblossomC130.html

 

[rb1957]

in my experience (commerical transports) most loads monitoring equipment measures the motion of the airplane (mostly nz, sometimes roll rates) at the cg of the airplane and you use your stress model of the airplane to infer stresses in the structure. the reason for such "crude" measurement is that nz captures the most significant part of fatigue damage for the general airframe (knowing that you're dealling with an unusual operation of the airplane i thought roll might be significant); and secondly, you are usually limited in interpreting strcutural stresses due to external motion of the 'plane. You might notice a significant difference in stress due to, say, flaps down; most systems allow you to time stamp discrete actions (WoW, flap deplyoment, ...) and if you gather the load data in terms of time (takes up more memory, but these days, who cares ?) instead of more memory efficient algorithms (like range-pair, or stress peaks, which are not time stamped.

In my experience trying to gather in-service stresses is pretty difficult. one difficulty is determining the zero correctly ... if you s/gauge the lower wing skin, how do you know either the load in the skin when you install the gauge (= gauge zero) so what is the difference between a zero s/gauge reading and the internal stress in the skin; or how do you unload the skin so that gauge zero = zero internal stress ? after this issue comes the operational problems 'bout keeping these fussy things working all the time. I have seen strain guages installed to measure wing moment, the plane had wing mounted gear, and the static 1g-on-ground stress plot indicted a wing station with near enough zero moment. I have strain gauges a wing strut, 'cause this could be removed from the plane (unloaed), strain gauges installed, and re-installed.

Once you have a stress history, the rest is easy ... miners rule for example, rainflow, range-pair ... there are many ways to infer the damge of a stress history.

Again, in my experience, gather the data for a pretty long time 2, maybe 5, years; to catch seasonal changes.

What you might do with todays tools is simulate some of the key maneouvers involved in the anticipated usage of the 'plane, particularly those not considered in the original design of the 'plane ... severe rolling, banking maneouvers, severe pull-out ... there may be static questions.

 

[davidjh]

Airborne sensors have improved dramatically in that last two decades. There are several, small, highly accurate 3-axis accelerometer modules available. I'd think that measuring "G's" at several places would be a good methodology to "infer" stresses. G's would also be a good measurment to use to present real time caution/warning alerts to the flight crew. I agree w/ rb1957, SG's requires continual "maintenance" and judgement to provide valid data. Try the "AGARD" series of papers for more backgound. Goog luck -you are into a very interesting project. PS: I'm NOT a accelerometer salesman.

 

[crackman]

ibushong

My company is currently performing all of the full airframe fatigue evaluations of the P2V (completed) and P3 aircraft used by the USFS in the firefighting role. There are primarily 3 purposes for recording data:

1. Validation of Loads
2. Mission Profile Definition
3. Load Histories

Strain gages are primarily used for validation of loads. Their location is selected based on where the VMTs of the wing/tail/fus are to be validated. Gages are not typically used for local stresses or developing fatigue spectra at critical locations since it would take many of them to capture local discontinuities. Of note, the gages for loads validation should always be placed in a location closest to a section neutral axis to avoid such influences as bending effects and away from any discontinuities such as cutouts (ie away from areas of high stress concentration). The actual locations for the gages (such as wing stations) are chosen by the structures engineers based on where the analytical loads have been developed and on the best location for correlation.

For mission profile definition, parameters such as airspeed, altitude, OEW, TOW, Fuel weight, flap angle, etc. are required. With respect to firefighting, a good amount of recorded data is required as the missions can vary significantly based on the tanker configuration and capability.

For load histories, accelerometers are used. These are generally used for recording repeated loads at the CG of the aircraft due to maneuvers, gusts, landing, taxi, etc. Of particular note with respect to the firefighting environment, actual flight testing is required to seperate the Nz due to the drop versus pilot induced maneuvers. Since airtankers typically drop large quantities of retardant, the aircraft experiences a large Nz due to drop which is not discernable in recorded data as the Nz data does not separate pilot induced Nzs and the drop Nz.

Hope this helps and good luck.

 

[rb1957]

crackman makes a good point about the change in airplane weight. measuring nz means you need the associated a/c weight. for a transport mid-mission weight is a resonable assumption, for a fire-bomber i think you'll need to know which peaks happened before the drop and which after ... recording a load history (nz peak and time) would give you this.

like crackman, I'm not a slaesman, but we've used SoMat systems successfully ... but there are several systems out there.

thinking about this for one more second ... to understand how it inspect a plane you need to know where to look and how often. I'm assuming you have access to the original stress analysis of the plane and it's mods and that this tells you the critical locations (particularly stress risers); it's possible that the mod didn't investigate fatigue as thoroughly as it should have so there might be some details there to look at.

I think the loads monitoring exercise is focussed on determining the severity of the new mission, compared with some baseline assumption. presumably the original airplane had a fatigue spectrum as the basis of it's analysis. And the loads you gather need to be those that you can use to develop stresses; ie nz is fairly obvious (you can easily develop airplane stresses for a 1g load), flap angle could have a significant effect on some locations (again, i think it's pretty easy to derive a 1g load with flaps deployed), roll rate could be significant but have you the internal loads associated with the maneouvre ...

stress = K*nz+L*nz(flap down)+M*roll rate ...

but i ramble !

 

[crackman]

As an explanation to some of the good points rb1957 pointed out, the following is a description of the approaches we took in evaluating aircraft used in firefighting which had previous fatigue damage due to past usage.

The analysis is performed in 2 phases. The first must determine how much fatigue life was expended during its original mission (ie military usage). We accomplished this by completely redeveloping the P2V and P3 airframe external fatigue loads and military mission profiles. A completely new stress, fatigue and DTA analysis was performed (over 80 locations analyzed). Then, the fatigue damage was determined for each aircraft based on the flight hours and cycles accumulated during military service. Phase 2 entailed developing the firefighting mission profiles and developing the airframe external fatigue loads, stresses, fatigue and DTA. This is then compared to the number of flight hours and cycles accumulated during the firefighting mission. Quite a bit of engineering was required but entirely do-able. The main key with the recorded firefighting data is that you need a sufficient amount of recorded data to predict the usage. Also, flight testing is a must to baseline the instrumentation and loads verification.

In my opinion and experience, a single (or even a couple) severity factors is not an accurate approach for missions as severe as firefighting. It may be possible to use (and I have seen this done) severity factors for changes in commerical or business jet usage (ie longer or shorter missions, different fuel weights such as added tanks for range or increase gross weights) but these changes are minor. Missions such as firefighting include items such as the drop phase, lots of maneuvers (much more frequent than commercial usage - very common to see loads as high as 75% of limit every few flights), numerous flight control deployments. In summary, enought critical areas need to be re-evaluated to capture all of the particular aspects of the firefighting mission. So, parameters such as these must be recorded in order to perform an accurate evaluation.