| dc.description.abstract | Airports that operate under cold weather conditions face major challenges in ensuring that runways, taxiways and aprons provide sufficient tire-pavement friction to the operating aircraft. This thesis is motivated by two practical problems: (1) maintaining or improving the pavement surface conditions in an, for airline companies, acceptable state and (2) accurately reporting the actual surface conditions to the relevant actors (pilots, air traffic control, winter maintenance services). The primary objective of this thesis is to broaden the general knowledge base of these problems. The work can be divided into a practical, a fundamental, and an applied part of the thesis.
The practical part includes a field study on how runway surface conditions change in time and the consequences for runway operability. Different situations were documented where the runway surface conditions changed due to snow fall, sand displacement by aircraft, ice deposition, snow compaction, and melting of the contamination layer. These cases highlighted two weaknesses in the current reporting system: (1) the constrained inspection frequency of the runway surface and (2) the limited possibilities to monitor the surface conditions while the runway is open for air traffic. The practical part also included field studies on a new sanding method, based on pre-wetting the sand with hot water. Practical experiences from maintenance personnel were collected, runway surface conditions were documented, and comments from pilots on the reported conditions were investigated. The method provides a solution for the problem that loose sand can be displaced or blown off the runway by the engine thrust of operating aircraft. In addition, the study highlighted some potential negative effects related to the sanding method. The high friction values that are typically measured on surfaces treated with warm pre-wetted sand can create a too optimistic picture of the prevailing conditions for aircrafts. Cases are documented where pilots faced worse conditions than they expected from the provided friction numbers. In 66 % of the cases there were clear indications available that the situation was not as good as suggested by the friction measurements. Another aspect is the risk of Foreign Object Damage (FOD). Maintenance personnel pointed out the importance of proper pavement cleaning prior to the sand application.
The fundamental part of the thesis focuses on the role of sand in the creation of tirepavement friction on iced surfaces. The presence of sand particles changes the interaction between the tire, the pavement, the contamination layer, and the atmosphere in which the interaction takes place. Hence, it changes the way friction is created. The interactions were studied on a macroscopic scale by observing tire tracks on sanded, iced runways and by aircraft braking experiments on ice treated with loose and warm, pre-wetted sand. These observations showed that loose sand particles, ones trapped between the tire and the ice, can slide together with the rubber tread and plough into the ice layer. Loose sand particles can pile-up in front of, and under, locked tires (full skid). Such tire lock-ups can occur, even though when aircraft are equipped with anti-skid braking systems because these systems become disabled below a certain threshold speed (ranging between 30 and 45 km/h, depending on the aircraft type). On freeze bonded sand (produced by the warm, pre-wetted sanding method), friction can be provided by both loose particles that plough into the ice and by particles that stay fixed to the ice and force the tire tread to deform.
The sand-ice and rubber-ice interactions were also studied on a microscopic scale by etching and replicating the ice surface. These observations revealed that the sliding friction process involved ice deformation in both cases. During rubber-ice sliding friction, the original crystal structure of the ice remained intact during the interaction. However, small scale ice deformation was evident by the formation of dislocations, aligned in rows along the sliding direction, and by the formation of small scale ploughing tracks. In the case of sand-ice interaction, the ploughing of sand particles was accompanied with the formation of cells within the original crystal structure of the ice. This re-crystallization was observed both in the laboratory and in the field. The rubber-ice and sand-ice sliding friction mechanisms were studied quantitatively by using a British Pendulum Tester in a cold laboratory experiment. It was found that the observed variability in friction measurements was significantly larger than the uncertainties introduced by the instrument itself. The variability may be caused by poorly controllable/reproducible, microscopic or nanoscopic surface properties of the ice and rubber. Rubber-ice interaction resulted in appreciable friction coefficients (0.5 ≥ μ ≥ 0.2) at ice temperatures below -5°C. However, it dropped significantly (down to μ = 0.05) over the whole tested temperature range to by the presence of little snow on the ice (less than 1 mm). It demonstrated that friction provided by rubber-ice interaction is very vulnerable to snow contamination. In contrast, sand-ice friction measurements did not show the dramatic drop in friction by the same amount of snow. Hence, the ploughing of relatively large sand particles provided a more robust mechanism of friction, compared to rubber-ice friction.
The applied part of the thesis comprises an exploratory study on a non-contacting measurement principle to quantify the amount and distribution of sand particles on a pavement. A static laboratory arrangement was build where sanded pavements were illuminated by a visible laser light source (wavelength: 635 nm) at different angles of incidence. The radiance from the illuminated area was recorded with a digital camera at different angles. The test matrix included dry and iced pavements and different sand application rates. A correlation between the total radiance and sand application rate was only found when there was negligible radiance from the pavement in the scene. The sand detection therefore required a distinction between radiance originating from the sand and the radiance originating from the pavement. However, due to the similarities in optical properties of the sand and the aggregates in the pavement (both originate from crushed rock) and the transparency of ice in the visible range, it seemed unlikely that the distinction can be made on the basis of radiance intensity. An alternative approach was investigated, based on triangulation. Image analysis techniques were used to define a region of interest where the radiance only originates from the sand. Within this region, individual sand particles can be identified and counted. The principle was developed theoretically for flat surfaces and adapted for application on rough surfaces of unknown topography. It was tested on a selected group of images, taken under favourable incidence and camera angles. The algorithm placed the region of interest reasonably well in all analyzed images, resulting in a rather conservative input in the subsequent analyses. The sand detection algorithm had a success rate between 63 and 100 %, depending on the surface contamination. The errors were mainly caused by not detecting particles that were located in the lower parts of the surface topography. Only few mistakes were made by incorrectly identifying particles. Hence, the number of detected particles was a conservative estimate of the actual number of particles located in the region. | nb_NO |
