article

Engineering Sciences Data Unit (ESDU)

Posted: 26 May 2009 | Mr Ken Balkwill, Consultant to Engineering Sciences Data Unit, London, United Kingdom | No comments yet

The role of the Engineering Sciences Data Unit (ESDU) in Modelling Performance of Aircraft Tyres on Contaminated Paved Surfaces.

The role of the Engineering Sciences Data Unit (ESDU) in Modelling Performance of Aircraft Tyres on Contaminated Paved Surfaces.

Aircraft Performance is directly affected by runway contaminants in two ways. There is an increase in decelerating force because contaminants impinge on the airframe and wheels. There is also a decrease in the available braking friction. As a result, the consequences are mixed.

  • During the take-off ground run the extra drag on the wheels reduces the capability of the aircraft to accelerate
  • Additional drag is also created by the tyre-spray impinging on the aircraft structure or creating additional skin friction as a result of wetting action on the aircraft
  • During the landing ground roll or when stopping as a result of a rejected take-off, reduced braking friction and increased drag act to offset the effects each with the other
  • Ground handling of the aircraft is adversely affected as a result of changes in the lateral forces available.

Research background

Aircraft operations from runways affected by natural precipitation are a perennial concern. Since the middle decades of the twentieth century, there have been numerous test programmes aimed at improving prediction methods in these conditions or to identify when such conditions exist. Important test programmes have been conducted in Europe, North America and Japan. What has been lacking, however, is a coherent attempt to synthesise the various sets of data in the form of a mathematical model to clarify the effects of the factors involved. During the last twenty years, there have been two major international projects related to the subject of contaminated runways.

  • The European Union sponsored ‘CONTAM RUNWAY’ and
  • The Canadian-sponsored Joint Winter Runway Friction Measurement Programme (JWRFMP)

CONTAM RUNWAY led to the collection of relevant (existing) test data and prediction methods from European civil airframe manufacturers. New tests, particularly on smaller types of aircraft, were conducted as well.

JWRFMP was an initiative supported by more than thirty organisations in twelve countries. Tests were conducted using various aircraft and runway friction measuring devices – with particular emphasis on winter-contaminated surfaces.

Involvement of ESDU

Since 1970 the ESDU Performance Committee has issued a number of Data Items on aircraft tyre braking and drag forces and by 1997 had acquired significant experience in modelling braking performance on both aircraft and ground-test machines operating on wet runways. This led to an invitation to assist in the analysis of the test data produced by JWRFMP. In addition to using those data, a feature of the task was to seek to exploit the significant body of test data that had been accumulated over the previous forty years so as to enable the derivation of a mathematical model applicable to the widest possible range of operating conditions.

In addition, work has continued at ESDU to refine and develop the scope of the modelling. Research is still in progress, with particular emphasis on runway drainage and the effect of runway texture on planing.

1. THE ESDU CONTAMINATED RUNWAY MODEL

Mathematical modelling

There are two elements into which the analytical work and the resulting mathematical model naturally separate.

  • First, there are retarding forces due to the tyre rolling and the additional drag that arises from the effects of a significant layer of contaminant. Note that separate treatments are needed for fluids (water or slush) and particulates (loose snow).
  • Then there are braking forces generated between tyre and surface. Initially dry surfaces are considered and the effects of contamination are built from that base. Note that separate treatments are needed for water, ice and compacted snow.

The relevant, separate elements can be combined, as necessary, to cope with a range of circumstances. One example considered here is braking in deep snow.

Aircraft braking in loose snow

This case demonstrates the effectiveness of the prediction method well. A Boeing 737-100 with idling engines was braked on a runway covered in six inches of loose snow. The braking-friction/snow-drag model has been coupled with the aerodynamic model for the aircraft, to show the relative sizes of all the major contributions to longitudinal accelerometer reading.

Thrust produces a small acceleration, whilst the aerodynamic drag is significant at the higher speeds. Rolling friction, whilst small, is sufficiently significant to merit inclusion in the calculation. Snow drag is significant but, even though the runway is completely blanketed with snow, the braking action contributes 15% of the aircraft weight to deceleration. Whilst this is a significant reduction on the braking contribution on a dry runway (torque-limited-braking friction is about 0.45) the effect of such a snow cover is not disastrous. More importantly, the model predicts the performance of the aircraft almost exactly by using average values of snow properties and reference coefficient of friction. Slip ratio has been deduced from the dry braking run shown in Figure 1 below. All-in-all, this case shows that with careful attention to system details, the friction and contaminant drag methods that have been developed are capable of predicting accurately the performance of aircraft braking systems, using no more information than is currently available. Of course, at low speed, when the anti-skid system is disengaged, it is not possible to predict decelerating force due to braking.

Figure 1: Effect of speed on deceleration due to aerodynamic, rolling and braking forces

Figure 1: Effect of speed on deceleration due to aerodynamic, rolling and braking forces

On using the model

Modelling is dependent on knowledge of eight independent variables. These are

  1. Depth of macro-texture of runway surface
  2. Depth of contaminant
  3. Density of contaminant
  4. Tyre forward speed
  5. Tyre inflation pressure – this is always used in the model as an absolute pressure
  6. Vertical load on tyre
  7. Nominal width of tyre
  8. Nominal diameter of tyre.

Of these, only the first three are related to the runway and its condition. All the other1 quantities are part of customary calculations of ground performance. Whilst it is not mentioned in the list, information relating to the operational characteristics of the aircraft anti-skid system is also needed. This information can usually be couched in terms of an effective slip ratio, that may vary with ground speed. If this is not readily available, then it can usually be deduced from analysing normal brake tests on dry or wet runways so that the anti-skid system can be emulated.

It is of particular note that only nominal dimensions for the tyres are needed. This is because use is made of the concept of reference dimensions in much the same way that such dimensions are used in customary aeronautical studies.

2, LIMITATIONS

Controllability

The ability to control an aircraft on a contaminated runway is a function of aircraft inertia and the forces acting on it with particular reference to crosswinds, asymmetric braking and the use of reverse thrust. Although the modelling currently incorporates lateral forces for freely rolling wheels, the effects of braking on yawed wheels is still being studied and has not yet been quantified.

Surfaces for which validation exists

The modelling has been shown to be valid for untreated surfaces. However, compacted snow that has been treated with sand – a situation commonly encountered in Scandinavian countries – has not been taken into account. Other (chemical) treatments of ice and snow are also outside the scope of the current modelling.

Surface uniformity

In using the methods assumptions have to be made as to the extent and uniformity of contaminated surfaces. Analysis and predictions can be done easily if it can be assumed that depth and density of contaminants are uniform. In many cases, it is sufficient to make appropriate (conservative) assumptions. At the time of writing, techniques to measure or deduce these parameters in real time are under investigation within the context of the ESDU modelling methods.

Probability assumptions

For some parts of the methods there is insufficient information in the public domain to enable the construction of a deterministic model. For example, loose snow is a descriptor that can have a variety of implications in terms of density and hence shear modulus. Within the methods, the shear modulus of snow is treated in statistical terms and the user is invited to select a level of probability for use in the calculation. In effect, this selection is equivalent to putting a desired level of conservatism on the result and ultimately, is a function of the degree of scatter in the parameter that is being used. The decision as to what level of conservatism should be selected rests with the user. However, where the resulting aircraft performance data are to be used as regulated information, the appropriate Authority may provide advice as to the required level of conservatism.

Furthermore, even though the modelling is sufficiently deterministic in most parts, the provision of regulatory data using the model needs to be coupled with properly established margins for safety. In customary practice these margins are often arbitrary because the processes of analysis and estimation have been shrouded in unquantified uncertainty. Using the ESDU methods, coupled with an appropriate appreciation of statistical methods, it is possible to analyse even limited test data so that uncertainty is properly quantified. Regulatory data can then be couched in the appropriate statistical terms. These issues are part and parcel of the research currently being undertaken by the ESDU Performance Committee.

Reference

1. If side forces are to be calculated then yaw angle is needed in addition to the other variables

Send this to a friend