Review and Improvement of Runway Friction and Aircraft Skid Resistance Regulation, Assessment and Management

Abstract

Runway skid resistance is crucial for the safety of aircrafts. Despite being internationally regulated, investigation reports published by the Australian Transport Safety Bureau and the US National Transportation Safety Board indicate that 4.9–22% of runway excursion accidents are related to insufficient friction, or to friction overestimation. Consequently, based on this review of friction physics, aircraft accident reports, international runway surface regulation, and aircraft braking performance regulation, it was concluded that significant improvement in the management of runway surface characteristics can be achieved. Areas for potential improvement in the current systems for aircraft skid resistance include gaps in the operational reporting of prevailing runway contamination, as well as friction and surface texture measurement and interpretation protocols. Furthermore, aircraft braking performance regulations are not related to actual runway surface friction levels, resulting in reportedly good runways being found to provide inadequate aircraft skid resistance in certain conditions. Recommendations include improvements in the management of runway friction and texture measurement and analysis during pavement design, and through the service life of the pavement surfaces. Finally, the basis of an improved international runway surface engineering design and management system is outlined. Recommendations can reduce the risk of aircraft skidding accidents in the future.

1. Introduction

The airport runway is a complex structure that provides strength, durability, and friction requirements during an aircraft operation [1]. Consequently, runway design, construction, and maintenance are strictly regulated to provide safety for an aircraft. At the same time, the pavement surface is the most important part of a runway, since it is affected by the highest stresses, and it provides the necessary friction [1]. Most of the large commercial airports use either cement concrete or asphalt as the runway surface layer [2]. These are durable materials, but in many cases the friction of the pavement surface must be managed to ensure it does not become insufficient. This is especially the case for densely graded asphalt mixtures because the nature macrotexture is relatively low and usually requires some improvement. For this purpose, pavement grooving is often used [3]. This is important for achieving the overall goal of the runway surface design, which is to make sure that aircraft operations are safe and efficient.

Nowadays, however, the runway is facing bigger challenges due to the rapid developments of commercial aircraft [4]. Since their first introduction in the early 1900s, aircrafts have become larger and heavier and particularly since WWII, aircraft wheel loads and tire pressures have increased significantly, as have aircraft length, wingspan, and wheel base/track. The number of large aircrafts has increased as well, due to the economic development around the world. The introduction of more demanding aircrafts sometimes triggers runway, taxiway, and apron pavement upgrades [5], and this affects not only safety, but also economic concerns, since the maintenance of bigger aircraft is more expensive as well as increased average runway occupancy time. The introduction of better runway surface design and maintenance techniques contributes to the necessary safety and economic efficiency of airport pavements for these more demanding aircrafts.

New dry surfaces usually provide necessary friction. However, texture deterioration and contamination reduce the friction over time [6]. Furthermore, runway contamination is inevitable, usually due to rubber build-up during the aircraft operation. Hence, it is important to be able to determine if contamination is reducing friction to an unacceptable level. Friction reduction is most critical during wet conditions, which greatly influences the risk of friction-related incidents. The main reason for this is the increased probability of hydroplaning due to a reduction of draining capability of the surface. If hydroplaning occurs, aircrafts lose the ability to decelerate and maneuver. This event can be unexpected during the heavy rain, or wet snow precipitation, and regular runway maintenance, especially rubber removal, helps to reduce this risk.

To identify when friction is compromised, international regulators provide standard testing methodologies to determine the actual friction and risk of skid-related accidents. The main methods recommended by the International Civil Aviation Organization (ICAO) are continuous friction measurement (CFME) using a partial-slip wheel fitted to a car, or a car-towed device, and macrotexture measurement [7]. These methods partially correspond to highway friction measurement techniques. However, the runway is affected by different conditions, due to the greater weight and speed of an aircraft, compared to the weight and speed of a cars and trucks. For this reason, the management of runway friction is more critical than for road pavements, and the improvement in runway friction assessment techniques is widely recognized [8].

The most recent and significant attempt to improve the runway friction regulation system was the introduction of the Global Reporting Format (GRF) [9]. The GRF is based on the real-time assessment and reporting of the prevailing skid resistance and braking capability of an aircraft during landing based on the actual runway condition and contemporary experience of pilots and runway operators during landing on that runway. The GRF runway friction assessment methodology is mostly based on the amount and type of contamination on a runway but also includes the limited possibility of a change of a report based on the actual friction measurement results. This methodology, however, was only introduced recently, and it still has the potential for improvement.

Today, efforts are underway to improve and update the testing and friction monitoring techniques for runway surfaces, as well as to provide high-friction surfaces for better performance and safer aircraft operations. To achieve this, a deep analysis of interactions between friction testing techniques, runway construction and maintenance methodologies, and the friction phenomenon itself are necessary. However, this is not a new issue, with runway friction being studied since the first half of the 20th century [10]. A contemporary review and analysis of those historical studies will now contribute to the improvement and updating of the current runway friction management regulations.

The aim of this review is to improve the understanding of current runway friction management practices. This is achieved by reviewing records of accidents related to skid resistance, friction-related research, and testing practices. The friction phenomenon and mechanisms of friction loss are also described in regard to aircraft landing. This review also includes an analysis of the friction-related international and state regulations. Finally, various improvements in runway friction management are recommended for future consideration.

2. Friction Phenomenon

Friction between the pavement and the tire provides the braking force and movement resistance of an aircraft. This is governed by the fundamental rule of friction force (F_fr), which is shown in Equation (1). However, it is important to understand the difference between rolling resistance and friction force, since rolling resistance depends on tire and pavement deformation in general, and friction force is based on the interaction between the tire and the pavement in the contact zone only, including adhesion and hysteresis effects [11].

$$F_{fr}=\mu N$$

(1)

where μ is the friction coefficient, a function of the tire and surface characteristics, and N is the normal force, usually the aircraft wheel load.

2.1. Rolling Resistance

Rolling resistance, which can also be referred to as rolling drag or rolling friction, is a force applied to the moving tire due to macro deformation of the tire and pavement [12]. In normal conditions, the rolling resistance of a tire depends on three factors: the energy loss of rubber and reinforcements due to cyclic deformation of the rolling tire; frictional energy in the contact area; and the air resistance of the tire, where energy loss of rubber and reinforcements generates most of the drag [13]. The determination of rolling resistance is a relatively trivial problem that can be solved using simple testing techniques or finite element modeling. Examples of rolling resistance calculations for road vehicle tires can be found in various studies [14,15]. Similarly, a rolling resistance evaluation for aircraft landing and take-off has also been performed. First, in a study performed by NASA in 1937 [10], the influence of different tire and wheel configurations on the rolling resistance of an aircraft, and the effect of rolling resistance during take-off were analyzed. It was concluded that the rolling resistance of a tire mostly depends on the properties of the pavement surface, such as the macrotexture and viscoelastic properties [16]. In the case of unsurfaced runways and runways covered with snow, rolling resistance can be high enough to affect aircraft landing and take-off and there are a number of techniques for the calculation of rolling resistance in snow based on testing data [17]. This reflects the importance of aircraft operations on temporary snow runways in the Artic and Antarctic regions [18].

As well as the surface characteristics, rolling resistance also depends on the ground speed of an aircraft. Figure 1 shows the influence of aircraft ground speed on the rolling resistance on unsurfaced airfields with different grass lengths [19]. Since most commercial airports have paved runways with concrete or asphalt surfaces, the rolling resistance is very small and its calculation is not important, especially in moderate and warm climates where thick snow contamination is unlikely to occur. Consequently, rolling resistance is not considered further in this review.

2.2. Sliding Friction

Sliding friction, which is often referred to simply as friction, on the other hand, is a more complex phenomenon. In general, there are two main processes for determining the friction coefficient during movement, which are adhesion and hysteresis. Adhesion is a molecular and mechanical interaction in the contact zone between tire and pavement, and hysteresis is a tire deformation in the contact zone that is different from rolling resistance in general. These two mechanisms of interaction are presented in Figure 2. The adhesive interaction between the rubber and surface is much less than the hysteresis grip, according to different theoretical studies [20,21,22].

Friction generally depends on the texture parameters and surfaces. In terms of road–pavement surface friction, texture can be divided into microtexture and macrotexture, based on the size or wavelength of the irregularities. The size of these irregularities is important for both friction estimation and friction measurement. The types of texture will be discussed in detail in the next sections.

The effectiveness of the prevailing friction coefficient can be significantly affected by surface conditions. Hydroplaning is a phenomenon where water on a wet surface is not displaced from the contact area between tire and pavement at a rate fast enough to allow the tire to make contact with the pavement surface over its complete footprint area [23]. Water pressure develops at the surface of the tire footprint, and on the pavement surface beneath the footprint, which originates from the effects of either fluid density or fluid viscosity, depending on the conditions. This has resulted in the classification of hydroplaning into two types, namely dynamic and viscous hydroplaning [24], which are described below. Both types of hydroplaning can exist simultaneously and have the same impact on the braking friction of the tire.

2.2.1. Viscous Hydroplaning

Viscous hydroplaning typically occurs on wet runways that have a smooth microtexture [24]. The microroughness has an amplitude ranging from 0.01 mm to 0.1 mm. The water film may have a thickness as little as 0.01 mm. The surface microtexture creates friction by providing a large number of sharp pointed projections that, when contacted by the tire tread, generate high but localized bearing pressures [25].

The theory of viscous hydroplaning was studied as early as the 1960s [24], where the contact area between tire rubber and the surface asperities was described as a water film with three regions, referred to as the inlet, outlet, and central region, according to the degree of tire movement (Figure 3). Due to the viscous flow of the liquid between the tire and the asperity, the highest pressure (p*) occurs in the central region. When the water pressure is equal to the pressure from the applied load, direct contact between asperity and rubber does not exist. In this case, if the water film thickness (h*) is greater than the size of the microasperities, a hydroplaning effect occurs. According to the calculations presented in Moor’s study [24], friction decreases by 98% during hydroplaning.

2.2.2. Dynamic Hydroplaning

In contrast to viscous hydroplaning, dynamic hydroplaning is the result of the hydrodynamic forces developed when a tire rolls over a water-covered surface [26]. It occurs when the amount of fluid encountered by the tire exceeds the combined drainage capacity of the tire tread pattern and the pavement macrotexture. If the aircraft speed is sufficiently high, the uplift force developed in the fluid film is comparable with the tire inflation pressure. This causes the tire surface to buckle and produce a large region of fluid capable of supporting the loaded tire. In these conditions, a loss of contact between the tire and the pavement occurs, resulting in an extremely low coefficient of friction [27].

Dynamic hydroplaning modeling has been studied in several studies [23]. A contact model between the tire and the surface, including water film between locked tires and the surface, has been developed [28]. The water film depth was calculated at different distances from the center line, as well as the friction coefficient at different groove depths and water film thickness conditions. It was found that the water film increased, as the distance to the center line increased, which can increase the safety risk of landing when an aircraft wanders away from the center of the runway. At the same time, surface groove depth was found to be important, due to its influence on the water film thickness and the hydrodynamic pressure under the tire.

In a similar study [29], a hydroplaning model was developed and was used to evaluate the minimal maintenance level of groove depth according to Federal Aviation Administration (FAA) regulations. According to the study, there is a high risk of hydroplaning when the groove width and depth are less than 3 mm. Below this threshold, the rate of deterioration of skid resistance increased rapidly, which verifies the appropriateness of the groove maintenance threshold set by the FAA regulations [30].

Hydroplaning speed also depends on the tire groove pattern. Modeling results show that tires with circumferential non-interconnected grooves are more susceptible to hydroplaning. It has been shown that the combination of circumferential and transverse tire groves provides the highest hydroplaning speed when other factors are constant [31]. The same results appear in the case of runway grooving, with transverse groves consistently producing higher hydroplaning speeds than longitudinal grooving [32]. Aircraft tires, however, can only use a longitudinal groove pattern due to intense wear during landing [33]. This means that the provision of surface macrotexture is more important for runways than it is for roads.

For practical hydroplaning speed estimation, Equation (2) is commonly used [34]. This equation was established by NASA in the 1960s, during tests on a flooded runway. Later, in the 1970s, NASA modified an equation after investigations of aircraft hydroplaning accidents, suggesting that the spin-up hydroplaning speed for a non-rotating tire, as occurs at aircraft touchdown, might be lower than implied by Equation (2). As a consequence, estimated hydroplaning speed was reduced by 15%, as detailed in Equation (3) [35,36].

Figure 3. Water film between tire rubber and the sinusoidal smooth asperity of the surface [24].

Figure 3. Water film between tire rubber and the sinusoidal smooth asperity of the surface [24].

$$v_P=9\sqrt{P}$$

(2)

$$v_P=7.7\sqrt{P}$$

(3)

where v_P is the ground speed in knots and P is the tire pressure in lb/in².

Equations (2) and (3), however, do not include water film thickness. It is well known that hydroplaning speed can change with a change in film thickness [37]. The analysis, however, is complex due to variation in surface height and the small size of irregularities in the geometry. Consequently, the tire does not fully come into contact over the entire surface of the pavement, because it is not flexible enough to conform to all the small depressions [38,39].

Loss of contact between some parts of the tire and the pavement surface occurs during normal braking in wet conditions as well [34]. In this case, water creates different zones between the tire and surface, as presented in Figure 4. In Zone A, the dynamic pressure of the water under the tire is higher than the tire pressure; this is the dynamic hydroplaning zone. The draping zone (Zone B) begins when the tire elements, having penetrated the water film, start to drape over the major asperities in the surface and make contact with the smaller asperities. The actual contact zone (Zone C) is the region where the tire elements, after draping, have reduced to an equilibrium position vertically on the surface. The length of this region depends on vehicle velocity, pavement texture, and water film thickness. Hydroplaning occurs when the actual contact zone disappears.

2.2.3. Reverted Rubber Skidding

Another process of friction loss is called reverted rubber skidding, or reverted rubber hydroplaning. Reverted rubber skidding occurs when the layer of rubber within the footprint melts due to friction, which results in an increase in the porosity of the tire and the trapping od overheated steam between the tire and the pavement surface [41]. Tire heating in the footprint contact patch causes the film of water to become a cushion of steam. The heat reverts the rubber to its uncured state (thus, reverted rubber) [42]. However, it is not clear what causes the low friction during this phenomenon, since attempts by NASA to replicate it in a laboratory experiment were unsuccessful [43]. Despite this, since 1968, NASA has reported numerous cases of overruns that were believed to be related to reverted rubber hydroplaning [44]. To prevent reverted rubber skidding, NASA recommended providing good pavement surface macrotexture, without contaminants, and improving pilot braking procedures [43,45]. In practice, reverted rubber skidding is often identified by white skid marks on the runway surface, due to the cleaning effect of the overheated and pressurized steam trapped between the tire and the pavement surface (Figure 5).

Reverted rubber skidding occurs less in modern aircrafts, due to the modern anti-skid braking systems, but it can still occur due to anti-skid system malfunction (Figure 6). In fact, most reports of reverted rubber skidding are associated with anti-skid system malfunctions, or emergency brake application, such as in the case of the EMB-145 aircraft runway overrun in Hanover (Germany) in 2005, or in the case of the Bae-146-200A aircraft overrun in Stord-Sørstokken Airport (Norway) in 2006 [46,47]. Due to these advances in aircraft technology, reverted rubber skidding is not considered in most modern runway friction studies, because it is not truly a hydroplaning event, and its presence during an accident can easily be detected by examining the tires for flat spots [48].

2.3. Summary

Friction is a complex phenomenon that describes the resistance force between tire and runway surface during the movement of an aircraft. The overall drag during the movement of an aircraft consists of sliding and rolling resistance. Rolling resistance has a negligible effect for uncontaminated and paved runway surfaces.

Sliding friction, or just friction, depends on two components: adhesion and hysteresis. Adhesion is a process of intermolecular bounding between a tire and a runway, and it requires a sufficient contact area to affect the sliding. Hysteresis, however, is based on the tire and pavement viscous deformation during the movement. Calculation of both of those components is a difficult task that cannot be achieved by practical means. Furthermore, sliding behavior completely changes in the presence of contamination on the surface, which includes water, ice, dust, dirt, etc. The most dangerous of those processes is aquaplaning or hydroplaning, which occurs when the contact area between the tire rubber and pavement is filled with liquid water or steam. Moreover, during normal braking, all of these processes partially occur in different zones between the tire and pavement.

In this context, the analysis of friction behavior during landing needs to be mostly empirical. It is reasonable to take into account real events of braking-related accidents to understand the most common risks of friction loss.

3. Review of Friction Related Accidents on a Runway

In overrun and veer-off accidents, aircrafts depart the runway surface and can be damaged or destroyed, meaning they are a significant safety concern. It has been reported that 15% of all global landing runway safety accidents from 2010 to 2014 happened due to surface contamination and poor braking action [50]. Overruns make up 31% of all accidents, while 20% are attributed to lateral veer-offs. According to the report, one of the likely scenarios leading to runway safety accidents is inadequate regulatory oversight and consequential undetected contamination of the runway surface, resulting in poor braking action. Similarly, according to a previous analysis of aircraft accidents [51], 7.8% of veer-offs during the landing and 8.0% of overruns between 2006 and 2017 occurred due to poor runway conditions such as contaminants and poor braking. Furthermore, another study found that most runway excursions in the Boeing database between 2003 and 2010, occurred after a stable approach and occurred due to runway contamination [52]. Consequently, friction-related aircraft accidents are a significant issue that have been occurring over a long period of time.

3.1. Australian Transport Safety Bureau

The Australian Transport Safety Bureau (ATSB), which provides independent investigation of transport accidents and safety occurrences, reported 45 investigations of runway excursions in Australia between 2003 and 2023 [53]. Of those 45 accidents, 10 (22.2%) were related to insufficient braking performance. In four cases, insufficient braking performance and an underestimation of landing distance were the main reasons for the incident on the paved runway (

Table A1. Runway excursion related to aircraft braking capability investigations by the Australian Transport Safety Bureau.

Number	Occurrence Date	Location	Aircraft	Occurrence Category	Highest Injury Level	Summary	Reference
1	6 April 2023	Abrolhos Rat Island	GA-8	Accident	None	During the landing, the aircraft floated significantly more than the pilot expected, with the aircraft touching down about 247 m beyond the threshold of a non-paved runway. The pilot recalled braking as soon as they touched down and then realized that the aircraft could not be stopped before the runway ended. The pilot overestimated the braking capability of the aircraft, probably due to fatigue. Overestimation of the braking capability of an aircraft on the pavement was not a contributing factor but increased the risk.	[176]
2	26 December 2021	East Wallabi Island	GA-8	Serious Incident	None	During the flare, the aircraft floated more than the pilot expected and touched down near the parking area about midway along the non-paved runway (about 350 m from the end of the runway). After touchdown, the pilot followed the operator’s normal practice of retracting the flaps and then applied normal braking. As the aircraft approached the end of the runway, the pilot realized an overrun was imminent and applied maximum braking. Despite this, the aircraft overran the runway by about 15 m. Overestimation of the braking capability of an aircraft on the pavement was not a contributing factor but increased the risk.	[177]
3	30 April 2014	13 km SW of Launceston	Cessna 172	Accident	Minor	The pilot was performing a landing on the ground runway and estimated the landing distance to be 440 m. Despite the fact that it was sufficient for the aircraft, the surface did not provide the deceleration he had expected. The surface was dumped due to rain, as it had rained the previous day.	[178]
4	19 February 2014	Lilydale (ALA)	Beech A36	Accident	None	During the landing, the approach was a bit unstable; however, the instructor suggested landing. The aircraft touched down about 200–300 m along the non-paved runway, and the aircraft did not decelerate after applying brakes. The wet runway did not provide sufficient braking performance, and the pilot underestimated the landing distance.	[179]
5	26 October 2012	Innamincka Township (ALA)	Piper PA-39	Accident	None	During the landing, the aircraft touched down about a quarter of the way along the runway. A gravel runway with loose stones did not provide sufficient braking capability, which resulted in an overrun.	[180]
6	25 July 2005	Nhill	Piper PA-31-350	Accident	Serious	During the take-off, the pilot experienced resistance to rearward movement of the control column of the aircraft and decided to abort take-off. However, despite the fact that this was an appropriate decision, the actual braking distance was longer, which resulted in an overrun.	[181]
7	23 May 2022	Chinchilla Airport	DHC-8	Incident	None	The DHC-8 aircraft experienced one engine failure and was unable to use the engines for stopping assistance. The pilots estimated the landing distance on a wet runway to be appropriate; however, the anti-skid system failed, which resulted in reduced braking effectiveness. The pilots decided to use reverse thrust; as a result, aircraft veered off the narrow runway.	[49]
8	9 January 2020	Newman Airport	Fokker F100	Incident	None	During the normal landing on a paved runway in wet conditions, the aircraft experienced aquaplaning, which resulted in a runway overrun. CFME results obtained during the airport testing showed a low friction level below the maintenance level and partially below the minimum friction level. However, the friction level for 100 m averages is above the minimum level.	[54]
9	23 January 2014	Archerfield Airport	Fairchild SA226	Serious Incident	None	During the landing in heavy rain, aircraft pilots attempted to stop the aircraft; however, due to aquaplaning, the aircraft veered off the runway, which resulted in the runway excursion. No information about the friction coefficient on the runway is available.	[182]
10	24 November 2010	Hobart Aerodrome	Boeing 737-8FE	Incident	None	During the normal landing in wet conditions, the aircraft experienced normal braking in the first half of the runway; however, in the last 1000 m of the runway, aircraft deceleration was low. During the inspection of the runway, rubber build-up was found. On the day of the incident, the crew of another aircraft also reported slipperiness on the runway.	[175]

). Analysis of these cases shows that one of the key factors in aircraft safety is a lack of understanding of the actual friction coefficient on the runway. In cases 1–7, friction coefficient and understanding of the actual braking capability of an aircraft were not the main contributing factors. However, it is clear that good friction and an understanding of actual runway conditions can help avoid incidents. In case 7, reverted rubber marks on the tire can be clearly seen (Figure 6). However, ATSB did not report it specifically. This condition occurred due to an anti-skid system malfunction on a wet runway.

Cases 8–10 (Table A1), however, show that poor friction on a runway can be the main contributing factor for runway excursion. In all of these cases, the final approach and landing were normal, and the aircraft braking system worked as usual. Unfortunately, only in case 8 were runway friction testing results presented. However, in case 10, rubber build-ups and poor grooving conditions were reported.

In cases 3, 4, and 6–10 (Table A1), the runway surface was damp or wet (Figure 7). Water reduces the friction coefficient on the runway, increasing the risk of runway excursion. To avoid this risk, surface condition monitoring should be performed before and during the landing. ATBS concluded in some cases that pilots’ unawareness of surface conditions also contributed to runway excursions.

The most detailed investigation was published for the runway excursion at Newman Airport [54]. As it was reported, the runway surface at Newman Airport did not provide sufficient water drainage, which resulted in puddle formation in the aircraft wheel paths (Figure 7). This increased the risk of slipperiness-related accidents. Furthermore, CFME results were regularly obtained, and the friction was lower than the maintenance level reported in the case of 100 m averages (Figure 8). At the same time, 10 m average results show friction levels below the minimum value (Figure 9). ATBS reported that the friction values recorded later were generally higher, which indicates that friction measurements can change during the year, which also affects the awareness and management of poor surface conditions.

Based on the above review of overrun and veer-off accidents in Australia reported by ATSB, it was concluded that there are a number of factors that can increase the risk of a runway excursion related to slipperiness:

• Lack of awareness of actual surface conditions;
• Water contamination of the runway and rains and showers before and during the landing;
• Poor runway maintenance;
• The absence of friction testing results, or inaccurate test results.

Based on the analysis of ATSB reports and recommendations provided by ATSB, recommendations for the reduction in safety risks related to skid resistance were proposed. According to investigations, it is possible to lower the risk of runway excursion related to slipperiness by providing:

• Regular monitoring of the friction coefficient of pavement;
• Regular runway maintenance, including rubber removal, regrooving, and resurfacing;
• Constant monitoring of surface and weather conditions.

3.2. National Transportation Safety Board (USA)

Aircraft accident data from the United States were also analyzed, based on the investigation reports provided by the National Transportation Safety Board (NTSB) [55]. All report runway excursions within the USA from 2014 to 2024 were considered. At least 75 out of 1800 (4.2%) runway excursions happened because of insufficient braking during landing or take-off (

Table A2. Runway excursions related to poor braking investigation reports in USA from 2014 to 2024.

No.	Occurrence Date	Location	Aircraft	Occurrence Category	Highest Injury Level	Summary
1	18 October 2023	Bandera, WA, USA	Cessna 172N	Accident	None	The pilot’s decision to land at a closed water-soaked grass runway which resulted in a runway excursion.
2	12 January 2023	Yelm, WA, USA	Beech 95-C55	Accident	None	The pilot’s failure to attain a proper touchdown point during landing with a tailwind which resulted in a runway over run. Contributing to the accident was a wet runway.
3	27 December 2022	Hawthorne, CA, USA	Embraer EMB-505	Accident	None	A runway overrun due to the pilot’s failure to stop the airplane as a result of diminished braking action due to a water-contaminated runway surface.
4	29 November 2022	Batesville, AR, USA	Learjet Inc. 45	Accident	Minor	The crew’s failure to execute a go-around during the unstable approach and long landing which resulted in a runway excursion.
5	9 September 2022	San Diego, CA, USA	Gates Lear Jet Corp. 36	Accident	Minor	A runway excursion due to hydroplaning on a contaminated runway.
6	20 August 2022	Donnelly, ID, USA	Cessna 182Q	Accident	None	The pilot’s improper decision to take-off from a wet grass runway which did not allow for adequate braking traction when the pilot aborted the take-off which resulted in a runway excursion.
7	6 June 2022	Nashville, AR, USA	Cirrus Design Corp. SR22	Accident	None	The pilot was not able to stop the airplane on a wet runway due to hydroplaning.
8	29 April 2022	Portland, OR, USA	Cessna 180K	Accident	None	The pilot’s failure to maintain directional control during landing on a wet surface with a tailwind which resulted in a runway excursion.
9	9 March 2022	Pittsburgh, PA, USA	Honda Jet HA-420	Accident	None	The flight crew’s continuation of an unstable approach which resulted in a long landing on a contaminated runway. Contributing to the outcome was the captain’s full application of the emergency brake which resulted in hydroplaning and a runway excursion.
10	18 February 2022	Detroit, MI, USA	Embraer EMB-500	Accident	None	The pilot’s decision to land on the contaminated runway with previous reports of unfavorable braking action.
11	7 February 2022	Crane Island, WA, USA	Cessna T207	Accident	None	The pilot’s failure to obtain a proper touch down location on a wet grass runway resulting in a runway overrun and impact with terrain.
12	3 September 2022	Naples, FL, USA	Raytheon 390	Accident	None	The pilot was unable to stop the airplane during the landing on a damp runway.
13	26 August 2021	Banner Elk, NC, USA	Embraer EMB-505	Accident	None	The pilot’s failure to achieve the approach criteria for the available runway landing distances published in the POH likely as a result of the steeper-than-normal approach and the required left turn on short final to avoid the terrain surrounding the airport. Contributing to the accident was a lower runway friction than that assumed by the airframe manufacturer and the tire cornering forces imparted during the landing roll which reduced the airplane’s reduced braking effectiveness which when combined with a high approach speed, increased the required stopping distance beyond the runway distance available. Also contributing to the accident was the operator’s lack of consideration of airport topography in its Destination Airport Analysis Program.
14	24 June 2021	St. Louis, MO, USA	Piper PA 46-350P	Accident	None	The pilot’s failure to maintain directional control during the landing roll with hydroplaning conditions.
15	14 March 2021	Port Aransas, TX, USA	Piper PA 46-350P	Accident	None	The pilot’s failure to maintain proper airspeed on approach and his attempt to land on a wet runway with insufficient runway remaining resulting in an overrun and loss of directional control.
16	16 December 2020	Jacksonville, FL, USA	Embraer EMB-500	Accident	None	The flight crew’s failure to apply maximum braking immediately upon touchdown which resulted in a runway excursion. Contributing to the accident was (1) the slightly excessive airspeed approach and (2) the flight crew’s decision to land on a wet runway during heavy rain with little margin between the unfactored landing distance required and the landing distance available.
17	13 December 2020	Leadville, CO, USA	Eclipse EA 500	Accident	None	The pilot’s failure to maintain proper control of the airplane which led to an unstabilized approach and a long landing on a runway contaminated with ice and patchy packed snow resulting in a runway excursion.
18	2 December 2020	Lufkin, TX, USA	Cessna 551	Accident	None	The pilot’s decision to land on a runway that did not provide enough length to stop the airplane given the wet surface conditions resulting in a runway excursion.
19	1 November 2020	Fernandina Beach, FL, USA	Raytheon 400A	Accident	Minor	The flight crew’s improper decision to land with a tailwind on a wet runway which resulted in a runway overrun. Contributing to the accident was the co-pilot’s early retraction of the speed brakes and the pilot’s decision to turn off the anti-skid system.
20	10 October 2020	Yakataga, AK, USA	Douglas C54	Accident	None	The pilots’ decision to land on a wet soft runway which resulted in the loss of braking action upon landing and a subsequent runway overrun and nose landing gear collapse.
21	3 August 2020	Kalispell, MT, USA	Cessna A185	Accident	None	The pilot’s excessive speed while landing on a wet grass runway which resulted in a runway excursion and impact with a fence.
22	15 July 2020	Washington, NC, USA	Air Tractor AT502	Accident	None	The pilot’s failure to maintain direction during landing on a wet grass runway.
23	14 July 2020	Fuquay Varina, NC, USA	Piper PA 22	Accident	None	The pilot’s failure to attain a proper approach speed which resulted in an overrun of the wet grass runway.
24	14 July 2020	New Carlisle, OH, USA	Waco YKC	Accident	None	The pilot’s failure to maintain a proper approach speed and descent path to the runway which resulted in excessive speed at touchdown and insufficient runway remaining to safely bring the airplane to a stop before the end of the runway. Contributing to the accident was the airplane’s diminished braking due to the dew-covered grass runway condition.
25	26 February 2020	Farmingdale, NJ, USA	Learjet 55	Accident	None	The flight crew’s failure to stop the airplane on the available runway which was wet and resulted in the airplane impacting a ditch.
26	12 February 2020	Broomfield, CO, USA	Piaggio P180	Accident	None	The pilot’s loss of directional control on the snow-covered runway which resulted in a runway excursion.
27	24 January 2020	San Juan, PR, USA	Gulfstream 150	Accident	None	The pilot’s loss of control during landing on a wet runway after encountering standing water.
28	20 January 2020	Sun River, OR, USA	Cessna 172	Accident	None	The pilot’s failure to maintain directional control during the landing roll on a runway with icy patches which resulted in impact with a snowbank and a subsequent nose-over.
29	4 January 2020	Morristown, NJ, USA	Beech 200	Accident	None	The airplane underwent hydroplaning while landing on a wet runway which degraded its braking capability and resulted in a runway overrun onto grass and mud and the nose landing gear collapsing. Contributing to the accident was the pilot’s improper decision to land the airplane until it was near the runway midpoint due to fog over the approach end of the runway.
30	11 November 2019	Chicago, IL, USA	Embraer EMB145	Accident	None	The flight crew’s inability to maintain the airplane on the runway centerline after touchdown due to the reduced braking action resulting from the deteriorating weather conditions which caused the airplane’s departure from the runway surface. Contributing to the accident was the delay in performing the runway assessment for undetermined reasons and failure to close the runway. Also contributing to the accident was the controller’s failure to advise the accident flight crew that braking action was no longer consistent with the previously published notice to air mission which described braking action as good across all three runway zones.
31	6 June 2019	Weiner, AR, USA	Piper PA24	Accident	None	The pilot’s failure to maintain directional control during the landing roll on a wet runway which resulted in a runway overrun and collision with a pole.
32	4 June 2019	Robertsdale, AL, USA	Cessna 182	Accident	Minor	The pilot’s failure to obtain the proper touchdown point while landing on a wet turf runway which resulted in a runway overrun.
33	3 May 2019	Jacksonville, FL, USA	Boeing 737	Accident	Minor	An extreme loss of braking friction due to heavy rain and the water depth on the ungrooved runway which resulted in viscous hydroplaning. Contributing to the accident was the operator’s inadequate guidance for evaluating runway braking conditions and conducting en route landing distance assessments. Contributing to the continuation of an unstabilized approach was (1) the captain’s plan continuation bias, increased workload due to the weather and performing check airman duties and (2) the first officer’s lack of experience.
34	15 February 2019	Montague, CA, USA	Cessna 402	Accident	None	The pilot’s failure to maintain directional control while landing on a wet/icy runway.
35	11 February 2019	Richmond, IN, USA	Beech 400	Accident	None	The flight crew’s decision to continue an unstable approach under conditions that exceeded the airplane’s landing performance capabilities which resulted in a runway overrun and impact with terrain.
36	1 February 2019	Appleton, MN, USA	Cessna 172	Accident	None	The student pilot’s loss of directional control while landing on an ice-covered runway which resulted in a loss of directional control and his failure to attain sufficient airspeed during a subsequent aborted landing which resulted in a nose-over.
37	6 December 2018	Burbank, CA, USA	Boeing 737	Incident	None	The flight crewmembers’ decision due to plan continuation bias to continue the approach despite indications of windshear and a higher-than-expected tailwind and the flight crew’s misperception of the airplane’s touchdown point which was farther down the runway than the crew assumed because of the faster-than-expected groundspeed. Contributing to the accident was Southwest Airlines’ lack of guidance to prompt flight crews to reassess operator-provided landing data when arrival weather conditions differ from those used in the original landing data calculation.
38	10 August 2018	Waynesville, OH, USA	Cessna U206	Accident	None	The pilot’s inability to stop the airplane on the wet runway after an aborted go-around which resulted in a runway overrun and impact with terrain.
39	3 August 2018	Greenville, SC, USA	Beech 58	Accident	None	The pilot’s failure to attain the proper touchdown point on a wet runway which resulted in insufficient runway remaining to safely stop the airplane.
40	19 July 2018	Oregon, WI, USA	Piper PA28	Accident	None	The pilot’s improper decision to take off with a known brake malfunction which resulted in a collision with a barn during landing on a wet runway.
41	20 May 2018	Marlborough, MA, USA	Diamond DA 40	Accident	None	The pilot’s improper decision to land on a runway that was too short for a safe landing and his subsequent failure to maintain directional control in tailwind conditions.
42	6 May 2018	Clarksville, TN, USA	Cessna182	Accident	None	The pilot’s decision to land with a tailwind in variable crosswind and downdraft and updraft conditions on a wet runway which resulted in a runway overrun.
43	5 May 2018	Bayou La Batre, AL, USA	Yakovlev YAK 52	Accident	None	The pilot’s improper planning for landing on a wet grass airstrip which resulted in a runway overrun.
44	12 March 2018	Madison, SD, USA	Cessna140	Accident	None	The flight instructor’s delayed aborted take-off on a soft wet grass runway which resulted in a runway overrun.
45	17 February 2018	Mount Sterling, KY, USA	Cessna402	Accident	None	The pilot’s failure to maintain directional control during the landing roll on a wet surface with reduced braking capability.
46	4 February 2018	Cleveland, OH, USA	Raytheon Aircraft Company 400A	Accident	None	The airplane’s reduced braking performance due to an ice-covered runway which resulted in a runway excursion. Contributing to the accident was the crew’s selection of a runway with a tailwind.
47	27 December 2017	Michigan City, IN, USA	Cessna 525A	Accident	Minor	The flight crew’s improper decision to land on a snow-covered runway that had insufficient runway distance for the airplane to land with the contamination which resulted in a runway overrun and impact with obstacles.
48	16 November 2017	Melborne, FL, USA	Piper PA 28R	Accident	None	The pilot’s loss of directional control during the landing roll on the wet runway with reduced braking capability.
49	4 September 2017	Lockwood, MO, USA	Cessna 182	Accident	None	The pilot’s unstabilized approach and failure to go around which resulted in a runway overrun on a wet grass runway.
50	23 July 2017	Wichita Falls, TX, USA	Beech G35	Accident	Serious	The flight’s encounter with adverse weather conditions at night which resulted in the pilot’s loss of airplane control upon landing due to standing water and a subsequent runway overrun.
51	17 April 2017	Indianola, IA, USA	Cheung David S Vans RV6	Accident	None	The pilot’s failure to maintain an adequate approach path which resulted in a long landing and subsequent runway overrun.
52	23 January 2017	Durango, CO, USA	Beech C 99	Accident	None	The pilot’s inability to maintain directional control during take-off in crosswind conditions on a contaminated runway which resulted in a runway excursion.
53	16 January 2017	Howell, MI, USA	Textron 525C	Accident	Serious	The pilot’s attempted landing on the ice-covered runway which resulted in a runway excursion and impact with terrain. Contributing to the accident was the airport personnel’s lack of training regarding issuance of NOTAMs.
54	23 December 2016	Warren, MN, USA	Cessna150	Accident	None	The noncertificated pilot’s failure to maintain directional control during the landing which resulted in a runway excursion.
55	4 October 2016	Portsmouth, NH, USA	Chaudoin George S RV 10	Accident	None	The pilot’s failure to go around after recognizing that the airplane was high and fast which resulted in a long landing on a wet runway and a runway excursion.
56	22 September 2016	San Juan, PR, USA	Learjet 25	Accident	None	The pilot’s improper decision to land the airplane on a wet runway in heavy rain with tires worn beyond safe limits which resulted in a hydroplaning condition and subsequent loss of directional control.
57	24 August 2016	Fallbrook, CA, USA	Cessna 182	Accident	Minor	The flight instructor’s failure to go-around and the subsequent long landing and his failure to maintain directional control which resulted in a runway excursion.
58	26 July 2016	Sugar Land, TX, USA	Embraer EMB-505	Accident	Minor	The airplane’s hydroplaning during the landing roll which resulted in a runway excursion. Contributing to the accident was the pilot’s continuation of an unstabilized approach, his decision to land in heavy rain conditions and his improper use of the main and emergency brake systems. Also contributing was the air traffic controller’s failure to disseminate current airport weather conditions to the flight crew in a timely manner.
59	21 July 2016	Baldwin, WI, USA	Cessna 208B	Accident	None	The pilot’s decision to land the fully loaded parachutist drop airplane on a wet grass runway that had insufficient length for the landing in high temperature conditions which resulted in a runway overrun when a more suitable longer runway was available at a nearby airport.
60	18 July 2016	Lenoir, NC, USA	Aeronca 7AC	Accident	None	The pilot’s encounter with instrument meteorological conditions which resulted in an emergency descent through fog a runway overrun on a wet grass runway and a collision with a ditch and vegetation. Contributing to the accident was the pilot’s failure to obtain a weather briefing prior to the flight.
61	2 July 2016	Hungry Horse, MT, USA	Cessna 182	Accident	Minor	The pilot’s decision to land on unsuitable wet terrain and his failure to stop prior to the end of the runway which resulted in a runway overrun and impact with tree(s).
62	4 June 2016	Williamstown, NJ, USA	Mooney M20C	Accident	None	The pilot’s landing area overshoot and failure to maintain surface speed and braking capability which resulted in a runway overrun and a collision with a fence.
63	22 May 2016	Griffin, GA, USA	Beech A36	Accident	None	The pilot’s decision to land with a tailwind resulting in a runway overrun and collision with a highway road embankment.
64	30 March 2016	Frenchville, ME, USA	Beech C23	Accident	None	The pilot’s failure to maintain directional control during the takeoff roll on a runway contaminated with ice and snow which resulted in a runway excursion and an impact with a snowbank.
65	25 January 2016	Lodi, CA, USA	Cessna 210	Accident	None	The pilot’s failure to attain a proper touchdown point which led to a runway excursion.
66	21 March 2015	Woodstock, GA, USA	Piper PA-28-235	Accident	None	The pilot’s failure to attain the proper touchdown point on the short, wet turf runway. Contributing to the accident was the pilot’s use of a high approach speed.
67	5 January 2015	Marquette, MI, USA	Cessna 172M	Accident	None	The pilot did not maintain directional control during takeoff on the snow- and ice-covered runway in gusting crosswind conditions.
68	21 November 2014	Sugarland, TX, USA	Embraer S.A. EMB-500	Accident	None	The pilot’s engagement of the emergency parking brake during the landing roll which decreased the airplane’s braking performance and prevented it from stopping on the available runway. Contributing to the pilot’s decision to engage the emergency parking brake was the expectation of a faster rate of deceleration and considerably shorter wet runway landing distance provided by the airplane flight manual than that experienced by the crew upon touchdown and an actual wet runway friction level lower than the assumed runway fiction level used in the calculation of the stopping distances published in the airplane flight manual.
69	20 November 2014	San Antonio, TX, USA	Beech 58P	Accident	Minor	The pilot’s improper decision to land long (past the midpoint) on a wet runway and his failure to conduct a go-around when the airplane did not touch down at the approach end of the runway which resulted in an overrun.
70	16 November 2014	New Buffalo, MN, USA	Cessna 150L	Accident	None	The student pilot’s failure to maintain directional control and the flight instructor’s delayed remedial action during landing. Contributing was the snow on the runway surface.
71	26 October 2024	Monument, OR, USA	Cessna TU206G	Accident	None	The loss of braking action during the landing roll on a gravel runway for reasons that could not be determined based on the available information.
72	19 September 2014	Leesburg, FL, USA	Cessna 172M	Accident	None	The flight instructor’s delayed remedial action and his subsequent loss of directional control during landing on a wet runway for reasons that could not be determined because post-accident examination of the airplane revealed no anomalies.
73	19 September 2014	Conroe, TX, USA	Embraer EMB 505	Accident	None	The second-in-command’s (SIC) engagement of the emergency parking brake (EPB) which decreased the airplane’s braking performance and prevented it from stopping on the available runway. Contributing to the SIC’s decision to engage the EPB was the lower-than-anticipated deceleration due to a wet-runway friction level that was far lower than the levels used to determine the wet-runway stopping distances in the Airplane Flight Manual (AFM) and it necessitated a landing distance considerably greater than that published in the AFM.
74	13 September 2014	Gaithersburg, MD, USA	Cessna T210N	Accident	None	The pilot’s failure to execute a missed approach after recognizing that the airplane was not aligned with the wet runway which resulted in a long landing with a tailwind and a subsequent runway overrun.
75	21 May 2014	Elkton, MD, USA	Cessna 182	Accident	None	The pilot’s failure to attain the proper touchdown point during a no-flap landing to a wet down-sloping runway and the airplane’s dynamic hydroplaning after touchdown which resulted in a runway overrun. Contributing to the accident was a total loss of electrical power due to an alternator failure.

). This number is lower than the numbers obtained by other researchers [50,51,52] and the number of occurrences based on the data provided by ATSB. However, in this study, the number of occurrences related to insufficient braking was likely underestimated due to the character of the NTSB reports, where information about braking efficiency during the landing and contamination is not provided in all reports. Thus, this analysis does not provide information about the occurrence frequency. Rather, it provides information about the causes of poor braking action. At the same time, it is impossible to provide any recommendations related to runway maintenance and runway friction control, since the reports provided by NTSB do not include such adequate information, like the details provided by ATSB.

Most of the occurrences reported for the United States (Table A2) had more than one contributing factor (Figure 10). Only 20% of all runway excursions due to poor braking had no other contributing factor(s). One of the most common contributing factors was a mistake during the approach and landing, which included excessive speed, long landings and unstable approaches to the runway. According to the study [56], pilot mistakes during approach and landing was the largest contributing factor to all runway excursion events in general. The second contributing factor was a lack of awareness of the prevailing runway conditions, which included either failure to update or provide a runway condition report, or ignoring the condition report by pilots. This also included an underestimation in the amount of contamination reported to the pilots. Wind-related problems included landing with a tailwind and gusting crosswind, with 15% of all runway excursion events due to poor braking being affected by wind conditions. In 11% of cases, the other contributing factors included incorrect brake application. In most of the cases, it was emergency brake application or the anti-skid braking system turning off that caused hydroplaning and the subsequent loss of directional control (cases 9, 19, 54, 58, 68, and 73 in Table A2).

Most of the occurrences in USA between 2014 and 2024 that related to insufficient braking happened during landing. Overall, at least 4.9% of runway excursions during landing were related to poor braking, and at least 1.3% of runway excursions during take-off were related to poor braking (

Table 1. Phase of flight during which a runway excursion occurs.

Description	Landing	Take-Off
Occurrences related to poor braking	70 (4.9%)	5 (1.3%)
Number of occurrences	1424	376

). Therefore, the landing phase is considered to be the most critical in regard to the risk of runway excursion events.

This analysis provides useful information to pilots and airport operators and allows them to minimize the risk of runway excursion by consistently avoiding the identified contributing factors in poor runway conditions with reduced friction. It is also reasonable to take measures to inform pilots about the risks related to poor braking. This analysis, however, does not allow for the calculation of the influence of poor braking on the risk of runway excursion, since it is impossible to obtain data about all of the runways with poor braking. That is because, in most cases, the poor braking condition was only reported after the incident, which does not allow comparison of the risk of runway excursion on runways with poor braking due to contamination or poor friction maintenance, to runways with good braking on a large sample. Perhaps it is possible that reported poor braking improves pilot awareness and decreases the risk of a runway excursion. Although the analysis of causes and processes of runway excursions clearly states that this is impossible, it is important to perform an additional analysis of statistics, based on the data from a number of airports with provided CFME testing results and meteorological archives.

In an attempt to assess the influence of low runway friction due to rain contamination, on the risk of runway excursion, seasonal aircraft incidents were correlated to average seasonal precipitation. Since precipitation affects runway friction, the correlation between the risk of a runway excursion and the relative amount of precipitation indicates the influence of low friction on the risk of a runway excursion. For the 3024 runway excursions included in the data provided by NTSB, each was grouped by state, and each season was valued by the relative number of incidents during this season for each state. Figure 11 shows the resulting correlation between runway excursion frequency and relative precipitation during each season. Relative precipitation was based on data collected by weather stations throughout each state from 1971 to 2000 [57]. The Pearson correlation coefficient was 0.23, which shows a weak correlation. That means that low friction increases the risk of runway excursion. However, the R-squared value is low since the analysis is generalized and not precise, but it can be used as qualitative evidence of such dependence. It is important to perform a more detailed study to assess the influence of low friction on the risk of a runway excursion, with prevailing conditions at the time of the incidents required, rather than seasonal averages.

3.3. Chinese Civil Aviation Administration

Similar results from aircraft accident data were recently obtained by Chinese researchers [58], based on runway excursion incidents in China during the period of 2013 to 2023. All of the runway excursion incidents were analyzed and 21 basic events were concluded to be critical. Of those, events associated with runway friction and braking force included deficiency in pilot techniques, ice and snow accumulation on runway surfaces, water accumulation on runway surfaces, and no grooving of runway surfaces. These are similar to the causes of failures identified in many Australian aircraft accidents.

The study used the fault tree analysis method for the analysis of those basic events, and critical influencing factors were captured [58]. The most critical influencing factor was concluded to be system failure; the next were deficiencies in pilot technique and violating standard operating procedures, ice and snow accumulation on airport runway pavements, and water accumulation on runway pavements.

3.4. Related Research

The risks of dangerous occurrences have also been studied by other researchers. A study [59] based on the same NTSB data shows that out of all meteorological conditions, wind and visibility most significantly increased the risk of runway excursion. According to a runway excursion risk assessment [60], in wet and standing water conditions, runway excursion risk is higher, and similar results were presented in another similar study [58]. Furthermore, a study presented a veer-off risk estimation method [61]. According to this study, wet runways increase the risk of veer-off by 50%, depending on the presence of other contributing factors. At the same time, there has been no recent study on the influence of poor braking action on the risk of a runway excursion. Finally, a report presented by the Fight Safety Foundation in 2009 [62] assessed worldwide runway excursion accidents from 1995 through to 2008. The authors concluded that runway contamination is a significant factor, which increases the risk of runway excursion and that at least 16% of all runway excursions included ineffective braking action as a factor [62].

It is important to mention that runway friction affects not only aircraft safety, but also airport efficiency (Figure 12). In a study performed at Vnukovo Airport (Moscow, Russia), it was found that runway friction affects the total taxiing and take-off time of an aircraft [63]. This would particularly be the case if friction meant the difference between regularly taking an earlier exit taxiway, or not. However, no similar studies have been found regarding the effect of runway friction on runway occupancy times.

3.5. Summary

In this section, an analysis of runway excursion incidents in Australia and the USA was performed and was compared to a similar analysis from China and to other studies based on investigations from other countries. It was concluded that poor friction significantly affects the risk of runway excursions. This conclusion, however, needs further consideration due to the complexity of the analyzed data. It was found that most of the runway excursion events, which happened due to poor braking, occurred during aircraft landing rather than take-off.

Figure 12. Total taxiing and take-off time of an aircraft and surface friction [63].

Figure 12. Total taxiing and take-off time of an aircraft and surface friction [63].

In all of the analyzed incidents, poor braking was the main reason for the runway excursion. In most cases, however, there were other contributing factors, which included:

• Approach and landing mistakes;
• Unawareness of runway conditions;
• Poor condition of the runway surface, which includes wear, rubber build-up etc.;
• Contamination of a runway with water, snow, ice, dew, etc.;
• Wind gusts, crosswind, and tailwind;
• Incorrect brake application and/or brake or anti-skid system malfunction.

It was concluded that it is possible to significantly reduce the risk of such accidents using the following recommendations:

• Regular monitoring of the friction coefficient of the pavement;
• Regular runway maintenance, including rubber removal, regrooving, and resurfacing;
• Constant monitoring of surface and weather conditions to increase the awareness of pilots.

Findings of other researchers in other countries are similar. Hence, it is important to analyze and improve international standards and regulations that will significantly reduce the risk of runway excursion due to poor braking.

4. International Runway Skid Resistance Standards and Regulations

The International Civil Aviation Organisation (ICAO) publishes a number of international standards regulating aerodrome and runway design and operation [64]. The main standard for aerodromes and runways is Annex 14 to the Convention on International Civil Aviation, commonly known as Annex 14. In terms of skid resistance and runway friction, there are five manuals to Annex 14, and the Procedures for Air Navigation Services (PANS)–Aerodromes (Doc 9981), and the associated Circular 355—Assessment, Measurement, and Reporting of Runway Surface Conditions, contain guidance on runway surface characteristics. All of these standards provide specifications, recommendations, and additional information about the requirement to provide and maintain surfaces with an acceptable level of skid resistance (Figure 13).

Two separate circumstances of aircraft skid resistance need to be considered: a wet runway, and a snow or ice-contaminated runway surface [65]. Wet runways are focused on in this review. To avoid potential problems caused by inadequate aircraft skid resistance provided to an aircraft, there are two possible approaches: the provision of reliable data related to skid resistance, in real time, to allow pilots to predict the actual braking performance of the aircraft, and the provision of adequate skid resistance at all times, under all reasonable and foreseeable environmental and operational conditions. In general, the second approach is implemented in most ICAO procedures and specifications, because the first approach is too difficult to manage and has the potential to create more risk. However, real-time contamination reporting is also implemented in the Runway Condition Report (RCR) system, which includes the runway condition code (RWYCC), based on the weather conditions, operator experience, and friction measurements [66].

Skid resistance is a complex phenomenon that is hard to measure directly. Basically, it is the maximum braking force that can be generated between a given tire, and a given surface, in given conditions. This significantly limits the possibility of precise friction measurements and the development of a unified minimum friction value. Therefore, the measurement of friction in controlled conditions is preferred, which can be compared to a reference number for an assessment of surface conditions. Besides that, indirect measurements, such as micro/macrotexture measurement, can also be used.

4.1. Wet Friction Measurement Results

According to the Manual on Certification of Aerodromes [67], airports should provide airfield pavement maintenance equipment, including friction measurement devices. Provision of sufficient friction is recommended to be controlled using continuous friction measurement devices (CFME) with a fixed slip ratio between 10% and 20%, measured immediately behind a 1 mm thick film of water applied to the surface. CFME is used during the construction of new pavements and periodically thereafter, to determine any deterioration in the friction level, either by surface ageing, surface erosion or surface contamination [68]. According to Annex 14 [7], the frequency of the runway friction survey should be sufficient to determine the trend of the surface friction characteristics of the runway. Doc 9137 P2 [34] suggests friction survey frequencies based on the operational tempo of the runway, as detailed in

Table 2. Friction survey frequency based on level of turbojet airplane operations for each runway end [34].

Number of Daily Minimum Turbojet Aircraft Landings per Runway End	Minimum Friction Survey Frequency
Less than 15	1 year
16 to 30	6 months
31 to 90	3 months
91 to 150	1 month
151 to 210	2 weeks
Greater than 210	1 week

, and most ICAO member states recommend these same frequencies. However, no testing frequency is provided for less busy runways, nor for runways with no jet aircraft operation. Some guidance is required for those less busy, but airports and their runway friction are still more important.

During CFME surveys, it is important to establish friction change dynamics for each runway, in order to plan maintenance (Figure 14). To avoid uncertainties in the measured values, it is important to identify anomalies, evaluate them, and reduce them where possible. Uncertainties can be grouped into the following categories [68]:

• Operator: anyone involved in the measurement process;
• Methods: specific requirements for performing the measurement;
• Means: any means (computer, software, device, and parts) used to perform the measurement;
• Materials: raw materials, such as tires, used to produce the final results;
• Environment: environmental conditions, such as temperature, location, and weather.

Reliable and repeatable operation of the friction measuring device can be improved through measurements of reference surfaces and constant control of results. Specific recommendations to increase the reliability of testing devices are provided. However, these do not include environmental conditions, such as temperature and humidity [68]. These factors could lead to errors and some researchers have recommended these factors be considered in runway friction measurement management [69,70].

There are a number of devices for friction assessment that can be used for conducting friction surveys. The list of devices and friction levels recommended by ICAO is presented in

Table 3. Runway surface condition levels [34].

Test Equipment	Test Speed, km/h	Design Objective for New Surface	Maintenance Planning Level	Minimum Friction Level
Mu-meter Trailer	65	0.72	0.52	0.42
	95	0.66	0.38	0.26
Skiddometer Trailer	65	0.82	0.60	0.50
	95	0.74	0.47	0.34
Surface Friction Tester Vehicle	65	0.82	0.60	0.50
	95	0.74	0.47	0.34
Runway Friction Tester Vehicle	65	0.82	0.60	0.50
	95	0.74	0.54	0.41
TATRA Friction Tester Vehicle	65	0.76	0.57	0.48
	95	0.67	0.52	0.42
RUNAR Trailer	65	0.69	0.52	0.45
	95	0.63	0.42	0.32
GRIPTESTER Trailer	65	0.74	0.53	0.43
	95	0.64	0.36	0.24

. Table 3 is based on the outcome of extensive trials performed at NASA’s Wallops Flight Facility in Virginia, USA in 1989. The various friction limits and targets are also reflected in many member state regulations, such as New Zealand [71], Australia [72] and the USA [30].

The reference friction measurement device for all CFME, dating from the 1970s, is the Mu-meter. ICAO Circular 355 states that the minimum friction levels from the table reflect historic levels for the individual friction measuring devices identified and have not adjusted to more recent comparisons of the devices. That is, the values in Table 3 have not been updated and reflect levels no longer considered unconditionally valid by ICAO [68]. Furthermore, Doc 9137 Part 2 states that the correlation between different friction measurement devices is not reliable under naturally wet conditions, and it is recommended to perform additional friction measurements under natural rain conditions, where friction characteristics can be reduced because of poor drainage due to inadequate slopes or depression [34]. It has also been stated by various researchers that those levels do not reflect the actual correlation between these devices [73].

The design objective levels listed in Table 3 are for information and reference only and are not for regulatory compliance of runways. These are meant to indicate the approximate values for a grooved surface, immediately after pavement construction and after resurfacing [34]. It has been demonstrated that the friction level of a newly constructed surface usually changes significantly within a relatively short period of time [74]. Hence, the only actual threshold for the friction level during the construction and early life of a pavement is the maintenance planning level, which provides a buffer to the actually important minimum value.

Friction levels measured by CFME (Table 3) are primarily intended for maintenance and resurfacing planning purposes [68]. They are not directly relatable to aircraft braking performance, which occurs at higher speeds, with different tire pressures and wheel loads, and in any prevailing condition of natural wetness. Consequently, CFME measurements should not be routinely reported to pilots for operational decision making. Rather, RCR should be used for operational management. The philosophy of RCR is that the aerodrome operator assesses the runway surface conditions whenever water, snow, slush, ice, or frost are present on an operational runway. This format of operational or real-time reporting was adopted by ICAO in 2021. Runway surface conditions are assessed and reported as the RWYCC. The RWYCC assessment and braking performance of an airplane are presented in

Table 4. Runway condition code assessment and braking performance of an airplane [66].

Runway Condition Code (RWYCC)	Runway Condition Description	Pilot Report of Runway Braking Action	Description of a Braking Action
6	Dry	N/A	N/A
5	Frost. Wet (≤3 mm) Slush (≤3 mm) Dry snow (≤3 mm) Wet snow (≤3 mm)	Good	Braking deceleration is normal for the wheel braking effort applied and directional control is normal
4	Compacted snow (outside air temperature below −15 °C)	Good to medium	Braking deceleration or directional control is between good and medium
3	Wet (slippery then wet due to poor micro or macrotexture) Dry snow (>3 mm) Wet snow (>3 mm) Dry snow on top of compacted snow Wet snow on top of compacted snow Compacted snow (outside air temperature above −15 °C)	Medium	Braking deceleration is noticeably reduced for the wheel braking effort applied or directional control is noticeably reduced
2	Standing water (>3 mm) Slush (>3 mm)	Medium to poor	Braking deceleration or directional control is between medium and poor
1	Ice	Poor	Braking deceleration is significantly reduced for the wheel braking effort applied or directional control is significantly reduced
0	Wet ice Water on top of compacted snow Dry or wet snow on top of ice	Less than poor	Braking deceleration is minimal to non-existent for the wheel braking effort applied or directional control is uncertain

. RWYCC 6 corresponds to normal braking in dry conditions; hence, no description of braking action is required. The assessment is performed for each third of the runway [66].

After an initial assessment, based on visual observation for contaminants on the runway surface, the RWYCC can be upgraded or downgraded based on observations, local knowledge and experience [68]. However, the process of downgrading and upgrading the RWYCC in wet conditions is not clearly explained, especially in the case of RWYCC 3, 4, and 5. Chapter 5 of Circular 355 [68] states that in the case of a wet runway with measured friction less than the minimum level, RWYCC 3 is appropriate, while Chapter 6 states that the repeatability, reproductivity, reliability, and friction levels for various devices are not adjusted according to a more recent comparison of these devices. This could lead to an overestimation of the actual braking capability for runways, with a friction level above the minimum friction level. For example, in the case of the runway overrun by the Fokker F100 at Newman Airport on 9 January 2020 (Table A1), the friction level on the runway was between the minimum and maintenance level; according to Circular 355, this corresponds to RWYCC 5, but the actual braking capability was lower than good. Furthermore, in the case of the RCR system, wet runways in warm climates can be attributed to either RWYCC 5 or RWYCC 3, without an intermediate grade. This is illogical and an intermediate rating should be introduced.

4.2. Micro and Macro Surface Texture

To maintain proper skid resistance, ICAO also recommends providing a good texture depth and performing rubber contamination removal and/or resurfacing as required [75]. The mean texture depth (MTD) of the runway surface should be more than 1 mm [7]. Proper texture enhances the contact between tire and pavement and increases the draining capability of the surface–tire contact area [7] and ICAO also recommends using surface materials with good texture, such as stone mastic asphalt, porous and open graded asphalt concrete. Alternatively, a surface treatment, such as grooving, is a recommended alternate [76].

ICAO also recommends providing good microtexture. However, there are no test methods for assessing the microtexture and microtexture limits are not specified. Rather, Doc 9981 [66] contains recommendations for microtexture assessment by touch and by visual assessment to detect excessive rubber contamination [66]. However, this is unreliable and is not recommended. Where microtexture is low, shot blasting can be used to increase the microtexture and to clean rubber contamination [68]. Ultra-high-pressure water blasting and chemical treatments are also available, with water treatments most common because they avoid the shots that can damage the aircraft if left on the surface, as well as the chemical treatments that can adversely impact the environment [77,78,79].

For macrotexture measurements, volumetric and profilometry methods are recommended [7]. The volumetric method evaluates the MTD value, while the profilometry method evaluates the mean profile depth (MPD) [80]. ICAO allows the use of both methods; however, volumetric methods, such as the grease smear and sand path methods, are preferred. Annex 14 [7] states that in the case of profile measuring techniques, a conversion between MPD and MTD must be established for the measuring equipment used [7]. At the same time, Circular 355 states that there is no universally agreed relationship between MTD and MPD [68].

4.3. Impact on Runway Length

During airport and runway construction and design, surface friction is not taken directly into consideration. The required runway length, in this case, mostly depends on the environmental parameters, such as temperature, slope, and elevation. The maintenance of standard or minimum levels of skid resistance is assumed in determining the required runway landing length. To ensure safe landing and aircraft operation, runway ends are usually connected with a stopway and a runway end safety area [67]. To avoid safety problems related to the aircraft overrun and to reduce the runway end safety area, an aircraft arresting system can also be installed. These areas are designed to decelerate an airplane overrunning the runway, by the use of high energy adsorbing materials of specific strength that will reliably and predictably crush under the weight of an aircraft [81]. However, these have many practical issues, such as their expense, reinstatement after use and access to the overrun aircraft to extinguish the fire and rescue the passengers [82,83].

As detailed further below, some high-friction surfaces, such as grooved surfaces and open graded surface mixtures, also allow a braking performance credit to be given to some aircraft [2]. By increasing the wet braking distance required for safe aircraft operations, effectively shortening the length of the runway required, it allows aircrafts to land on runways that would otherwise be considered too short. This is particularly beneficial for airports that were designed and constructed for smaller regional aircraft types but now support B737/A320 sized jet aircraft types.

4.4. Summary

International regulators provide recommendations for the control of runway surface friction. These regulations are based on the various parameters monitored for the provision of the necessary friction level. These parameters include surface friction using CFME, surface macrotexture, general groove and surface conditions, and contamination monitoring. The recommended testing procedures, however, are stochastic. Currently, there are no reliable characteristic values, and current testing methods need improvements, along with the development of standard protocols for the consistent and reliable determination of a characteristic value from the variable data collected. According to the previous section and based on the analysis of that section, it is clear that the current monitoring systems’ concept can lead to the risk of runway excursion, in particular:

• Friction assessment consistency and reliability cannot be ensured by the current requirements;
• Recommended friction levels are outdated;
• GRF does not allow adjustment of the RWYCC to intermediate friction levels based on the CFME tests;
• The influence of speed on friction is not reflected in the standards;
• There is no method for the calculation of a characteristic value based on the CFME results;
• There is no methodology for the assessment of the reliability of results for both texture measurement and friction measurement;
• Some of the values reflected in the standards do not include nominated test methods, for example, microtexture measurement, which is complicated and challenging.

It is clear that updating the regulations and guidance for runway friction specification, measurement and management would be beneficial and could contribute to reduced aircraft skid-related accidents. However, it is also important to link the physical runway characteristics to the assumed and regulated performance of the aircraft during braking on landing and during aborted take-off operations.

5. Aircraft Performance Regulations

In terms of aircraft performance regulations applicable to aircraft braking performance, ICAO has published two Annexes to the Convention on International Civil Aviation, and one manual (Figure 15). According to Part I and Part II of Annex 6 [84,85], pilots must be satisfied with the runway surface conditions available so that the aircraft can reliably and safely land and come to a stop within the landing distance available (LDA). To that end, runway length requirements should be included in the aircraft operations manual. According to the Annex 8 [86], landing performance data should be assessed at the time of take-off and landing. Doc 10064 [87] provides details for landing distance assessment, RWYCC interpretation, landing on contaminated runway, and general information on landing performance. During the landing performance assessment, pilots also need to consider information about the braking and landing capability of an airplane provided by the manufacturer.

5.1. Contaminated Runway Regulations

Recommendations for the effect of contaminants on runways is presented in the Aeroplane Performance Manual [87], based on the European Union Aviation Safety Agency (EASA) aircraft braking performance models [88]. These models provide levels of assumed surface friction coefficient available to the aircraft tire during landing on contaminated runways for different anti-skid braking systems (

Table 5. Wheel braking coefficients for RWYCC numbers [88].

RWYCC	Wheel Braking Coefficient
6	90 per cent of certified value
5	According to Figure 16 for a smooth (ungrooved and not textured) wet runway
4	0.20
3	0.16
2	50 per cent of the value obtained for RWYCC 5 for speeds below 85 per cent of the aquaplaning speed and 0.05 for speeds above 85 per cent of the aquaplaning speed
1	0.07
0	Not applicable

). RWYCC 0 corresponds to minimal or non-existent braking action; hence, no wheel braking coefficient in those conditions is applicable. The wheel braking coefficients presented in Table 5 correspond to the pilot report of runway braking action in Table 4. These values, however, are not related to particular aircraft models or aircraft and landing parameters.

Braking coefficients for RWYCC 5 are presented in Figure 16. Equations from [88] and [87] are presented graphically for simplicity and comparison. The assumed braking coefficient also depends on the tire pressure [88]. To determine the braking coefficient for a specific RWYCC and aircraft ground speed, the maximum braking coefficient is multiplied by the efficiency number of the anti-skid system, which can be defined during the anti-braking system tests. According to EASA, there are two methods for the evaluation of the anti-skid system efficiency value: the torque method and the wheel-slip method [88]. Both of these methods should be performed on a wet smooth runway, with worn tires, and the highest reasonable landing tire pressure, at different speeds and under different brake energy conditions. The approximate efficiency values for different types of anti-skid systems according to ICAO are presented in

Table 6. Efficiency values of anti-skid systems [87].

Type of Anti-Skid System	Efficiency Value
On-off	0.30
Quasi-modulating	0.50
Fully modulating	0.80

.

The on-off systems are the simplest types of anti-skid systems. For these systems, the brake is applied, and the pressure then releases when the wheel is locked. After the wheel accelerates back to synchronous speed, the full pressure is applied again [89]. This cycle is repeated throughout the stop, or until the wheel ceases to skid with the applied brake pressure. Quasi-modulating systems are an improvement of on-off systems and regulate the brake pressure continuously, as a function of wheel speed [89]. Typically, brake pressure is released when the wheel deceleration rate exceeds a preselected value. Brake pressure is reapplied at a lower level after a length of time, appropriate to the severity of the skid. Brake pressure is then gradually increased until another incipient skid condition is sensed. In general, the corrective actions taken by these systems to exit the skid condition are based on a pre-programmed sequence, rather than the wheel speed time history. Fully modulating systems are a further refinement of the quasi-modulating systems [89]. The major difference between these two types of anti-skid systems is in the implementation of the skid control logic. During a skid, corrective action is based on the sensed wheel speed signal rather than a preprogrammed response. Specifically, the amount of pressure reduction or reapplication is based on the rate at which the wheel is going into or recovering from a skid. Also, higher fidelity transducers and upgraded control systems are used, which respond more quickly. The type of anti-skid system and efficiency of the anti-skid system affects the maximum possible friction level and stability of braking. This effect, however, should not prevail over the overrun risk assessment, since the efficiency of the anti-skid system can change during landing, due to system malfunction, which was reported several times after the runway excursion (Table A1 and Table A2).

According to the international standards [87], the required landing distance for a contaminated runway can be defined by multiplying the landing distance from the flight manual by a landing distance factor (

Table 7. Landing distance factors.

RWYCC	6	5	4	3	2	1
Turbojet, no reverse	1.67	2.6	2.8	3.2	4.0	5.1
Turbojet, with reverse	1.67	2.2	2.3	2.5	2.9	3.4
Turboprop	1.67	2.0	2.2	2.4	2.7	2.9

). If the landing distances determined in the aircraft flight manual are presented as factored landing distances, then those values must be adjusted to remove the applicable dispatch factors applied. Landing distance factors published by ICAO correspond to factors published by FAA [90]. Furthermore, RWYCC 0 is not presented in the table since landing in those conditions is not advised [68]. According to ICAO [87], a full stop landing must be achieved within 60 per cent of the LDA for turbojet-powered airplanes, and 70 per cent of the LDA for turbo-propeller airplanes.

5.2. High-Friction Surface Treatment

Grooved and open graded friction course surfaces are deemed to be high-friction surfaces, primarily for the ability to positively drain the tire–pavement interface [87]. Although EASA allows an aircraft performance credit when operating on runway surfaces with a high-friction surface (Figure 16), ICAO does not. Additionally, the minimum friction thresholds that are required to be maintained by airports are no different for these surfaces than they are for smooth runways. Consequently, there are no provisions in place to ensure that the performance credit is actually maintained over time, even when the aircraft operator relies on them. This mostly affects airports with short runways that support larger, narrow-bodied transport aircrafts, such as the B737 and A320.

In general, aircraft performance requirements defined by ICAO and member ctates do not provide any methodology for considering the friction coefficient obtained from CFME surveys, which are commonly used for runway regulatory compliance. The only exception is for RWYCC downgrading of contaminated runways [66]. The Airport Services Manual for pavement surface conditions [34], however, provides a wet/dry braked stopping distance ratio (SDR) calculation, using results from runway CFME measurements. This method is based on the MU-EFF value, which is an effective friction value based on the testing device, brake application speed and airplane characteristics. Using the MU-EFF value obtained for a particular airplane, speed, and runway, it is possible to find the SDR using graphs for three-engine and two-engine jet airplanes (Figure 17). This method is based on the 1970’s research by NASA [45] but this method is no longer commonly used in practice.

5.3. Summary

International regulation of aircraft performance does not include any specific instruction or recommendations for the assessment of landing distance. Landing distance, however, should be determined for a particular aircraft based on the landing performance data and aircraft operations manual by the manufacturer. Hence, approximate data are provided to the pilot for consideration before landing. Elsewhere, the ICAO Aircraft Performance Manual contains recommendations based on the European and American aircraft certifications standards. These recommendations include landing distance and friction coefficient assessment, based on the RWYCC. These methods, however, do not take the airplane characteristics and CFME results into consideration. Part 2 of the Airport Service Manual, however, provides one method for the assessment of approximate landing distance based on the airplane type and friction measurement results. However, this approach is outdated and is no longer used in practice.

According to the results of the aircraft performance standards analysis, the following conclusions were obtained:

• Currently, there is no reliable method for the assessment of aircraft landing distance based on friction measurements and aircraft parameters;
• International standards do not provide a unified landing distance assessment methodology;
• It is important to upgrade the current landing distance assessment methodology to avoid risks of runway excursion.

It is also important to relate aircraft performance to the friction assessment and unify the friction assessment methodology. This highlights the importance of developing best practice surface characteristic assessment methods and consistent stochastic data analysis to determine a characteristic value.

6. Surface Characteristics Assessment

Friction characterization methods can be categorized using a system developed by Niu et al. [69]. As shown in Figure 18, these categories include fixed tests (A), braking tests (B), contact tests (C) and non-contact tests (D). Category A and category B tests measure the friction of the surface, while category C and category D tests measure the surface texture. The fixed tests (continuous friction measurement tests) are the most popular, in both highway and runway assessment, where A1 is the variable slip test, A2 is the side force test (lateral force test), A3 is the fixed slip test, and A4 is the locked wheel test. Braking tests are the simplest tests, especially the stopping distance test (B1), while the deceleration rate test (B2) requires additional equipment. ICAO regulations are based on CFME, which are fixed test methods, but it also recommends braking distance test methods when no CFME is available [34].

The contact tests presented in the figure are the British Pendulum Test (C1), the Dynamic Friction Tester (C2), the sand patch test (C3), and the outflow test (C4). Non-contact tests include the electro-optic test (D1) and the circular texture meter (D2). Although more test methods exist nowadays, this classification allows us to characterize the main principles of each test type.

6.1. Fixed and Braking Test Methods

CFMEs, also known as continuous friction measuring devices (CFMDs), measure the friction coefficient continuously while the testing device is moving along the pavement surface. Various devices are generally either towed trailers or modified cars. These devices have been used since the 1950s in road and highway friction evaluation and were introduced to the runway skid resistance assessment in the 1960s [91]. The first CFMEs were used widely in the USA, for example, such as the Penn State Road Friction Tester [92,93].

In 1990, NASA and FAA returned their attention to the CFME assessment of runway skid resistance and carried out a comparative study of different CFMEs during the FAA/NASA Runway Friction Program [94]. Tests were performed in different conditions, including dry runway, wet runway, ice contamination, and snow contamination, with six ground-testing vehicles and two instrumented aircraft. According to the results, ground vehicles correlated well to aircraft braking friction. However, the correlation did not allow the use of a CFME-measured friction value to be used for aircraft braking performance evaluation. All ground vehicle tests reported a friction coefficient that was significantly higher than the actual aircraft-measured friction coefficient [94].

Some authors have reported that the repeatability of CFME can be low [95,96,97,98]. For example, a comparison of the same CFME models showed 10–20% variability between different testing machines of the same type [97]. This means that the variation in results can be higher than the difference between the acceptable and minimum levels of friction on the runway. According to a similar study, the variation of the Grip Tester results depending on the tire actual diameter can be calculated using Equation (4) [99]. Based on this, it is clear that as the tire wears, the friction value goes up. Due to the allowed wear level of a testing machine tire, the value can increase by 0.02–0.05.

$$F=GN{\displaystyle \frac{\left(SD-CD\right)}{\left(MD-SD\right)}}$$

(4)

where F is a friction value, GN is a Grip Tester number, SD is a standard tire diameter (260 mm), CD is a chain cog effective diameter (130 mm), and MD is a measured tire diameter.

Moreover, as part of the research project sponsored by EASA in 2008, it was found that the braking slip, tire pressure, tire design, tire trad materials, derivation of friction coefficient, and self-wetting system all significantly affect the friction reading from different devices [73]. This part of the project was the harmonization of friction measurement techniques; as a result, basing the harmonization efforts on relationships to aircraft performance was considered to be the best option, which resulted in the development of the Global Reporting Format [100]. Friction measuring device harmonization, however, was not carried out due to the inconsistency of correlation models among devices. At the same time, it was recommended to use a Dynamic Friction Tester (DFT) as a reference device for the friction testing and a Circular Track (CT) meter as a reference device for the macrotexture measurement. These recommendations, however, were not incorporated into the ICAO standards [100].

According to NASA Annual Tire/Friction Workshop results, CFME reliability and repeatability can be increased by the following [101]:

• Performing the test during naturally dry surface conditions.
• Ensuring the self-wetting system accurately provides a reliable 1 mm water film immediately in front of the test tire.
• Performing testing at three test speeds, when assessing aquaplaning potential.
• Omitting the first two runs because these are often outliers to the rest of the results.

Some of the above-mentioned recommendations were presented in national standards, such as in Australia, New Zealand, the USA, etc. [71,102,103]. However, most of these recommendations only contain recommendations for surface condition controls, self-wetting system accuracy and general requirements, without specific procedures, especially when interpretating the outlying results. International standards have a similar issue [68]; hence, it should be recommended to incorporate the above-mentioned recommendations.

In the 1990s, the World Road Association, known as PIARC, performed an assessment of different friction testing techniques and developed the International Friction Index (IFI) [104]. The study, finalized in 1995, consisted of an investigation of friction and pavement texture testing methods. During the investigation, a correlation between 74 devices representing sixteen countries was found, and the Golden Value Friction Number (GF 60) and Golden Value Speed Number (GS) were evaluated. The correlation between each device and the golden number of that parameter was reported, and methods of correlation investigated for other devices were presented [104]. In a similar study, the International Friction Index for various devices was compared to the PIARC results and it was concluded that the correlation coefficients obtained by PIARC did not match their experiments [105]. The method of calculation of the predicted Golden Speed Value (S_p) also does not correlate with the experimental data. Furthermore, it was also concluded by others that the PIARC method did not correlate with new and independent experimental data [106].

Despite these shortcomings, the data obtained by PIARC can be used for friction assessment and comparison of different devices [107]. However, the factors that affect the results must be considered when comparing the PIARC results to other CFME trial results.

Influence of Weather Conditions on Friction Measurements

Weather conditions, such as temperature and humidity, can significantly affect runway friction [108]. Consequently, measured friction values differ from day to night, as well as from season to season, and climate to climate. This was confirmed by long-term analysis of CFME results performed in Australia [97], as shown in (Figure 19). The summer (December) values were generally 7% lower than the winter (June) values.

Researchers from New Zealand performed a similar analysis of data collected on highways, to develop a correction equation for the weather conditions [70]. The result was a model for estimating the yearly seasonal fluctuations in skid resistance, specific to the climate in New Zealand.

$$BPN=BP{N}_{terminal}-5\mathit{cos}\left(2\pi ∕365.25\times JDay\right)$$

(5)

$$GN=G{N}_{terminal}+0.002\mathit{cos}\left(2\pi ∕365.25\times JDay\right)$$

(6)

where BPN and GN are British Pendulum Number and Grip Number and JDay is a Julian calendar year. Equation (6), however, according to the authors, needs to be treated with caution as it is derived from data covering only one climatic region and aggregate type.

The influence of temperature on friction measurements for the British Pendulum and Grip Tester is shown in Figure 20 and Figure 21. Reported temperature sensitivity for the Grip Tester is similar to temperature sensitivity obtained by other researchers [109].

Similar results were obtained in another study, where both surface and water temperature influences on a friction coefficient measured with a Dynamic Friction Tester were analyzed [110]. According to the results, temperature can affect the friction coefficient significantly (Figure 22).

In all of the presented studies, high temperature reduced the measured friction value. One of the possible reasons for this effect is an increase in the wetting ability of the liquid with increased temperature. That assumption, however, needs to be studied further.

6.2. Surface Texture Assessment

Surface texture measurement can be used as an indirect tool for friction assessment and includes both contact and non-contact methods. Depending on scale, surface texture is categorized into four classes: microtexture, macrotexture, megatexture, and unevenness [111]. Figure 23 presents the wavelengths of texture classes and their influence on rolling tires, according to PIARC, with microtexture and macrotexture directly relating to aircraft skid resistance during wet surface conditions. Furthermore, unevenness and megatexture are undesirable and limited, due to other negative effects during landing and take-off on aircrafts, so only microtexture and macrotexture assessment and control are required for aircraft skid resistance and runway surface regulation.

Different authorities use different limits for the definition of microtexture and macrotexture. For example, ICAO [7] states that the threshold between microtexture and macrotexture is 0.1 mm, whereas PIARC reports 0.5 mm wavelength to be a threshold [112]. In general, microtexture can be described as a hardly detectable texture of individual stones, and macrotexture is a texture among individual stones that can be judged approximately by the naked eye. In practical measurements, the difference between macrotexture and microtexture is evaluated by using different algorithms, such as the Fourier transformation, the Butterworth filter [113], the Gaussian smoothing filter [114], or other algorithms, to effectively isolate the microtexture from the broader macrotexture.

In a state-of-the-art review [112], surface texture parameters were separated into two-dimensional amplitude and shape-related parameters, three-dimensional amplitude and shape-related parameters, spectral characteristics of a surface wave, and fractal and multifractal characteristics. Amplitude and shape-related parameters are widely used during the texture assessment. However, the other parameters have limited use. In one study [115], macrotexture, obtained by a circular track meter, was characterized using wavelet analysis, which allowed the analysis of the influence of different aggregate sizes on surface texture. In a similar study [116], asphalt surface friction was characterized as a set of fractal parameters of texture. Fractal mathematics was also used for the texture analysis by Kikkalis et al. [117]. The authors report that this technique has potential for a skid resistance assessment, but this has not yet been introduced into common practice.

Figure 24 presents methods currently used for texture assessment, with the methods recommended by ICAO shown in red. The methods are divided by measurement principle and purpose. Tactile and visual assessments can only be used for approximate texture monitoring. Visual assessment can be used for the detection of runway wear [118], rubber build-up [119], and pavement deformation [120], but not for direct friction characterization [66]. This type of assessment allows for the identification of areas with potential microtexture and macrotexture deterioration, but it is not a quantitative nor a precise assessment.

6.2.1. Volumetric Methods

Volumetric methods can be used for macrotexture assessment but are generally not precise enough for microtexture measurement. These methods are simple and generally do not require any special equipment. Consequently, they are recommended as the main method for texture assessment of runway surfaces. For the volumetric method, sand [121] or grease [91] are usually used. The average macrotexture is calculated by dividing the volume of sand or grease by the surface area of the sand patch or grease smear on the pavement surface. The resulting number is often referred to as the mean texture depth (MTD) [122].

6.2.2. Non-Contact Methods

Non-contact methods can be used for the evaluation of the macrotexture of pavement surfaces and some methods have greater precision than others, meaning they can also be used for microtexture measurement. These methods include image texture analysis [123,124], stereoscopy [125], computer tomography (CT) scanning [126], three-dimensional (3D) scanning [127], and laser profilometry [98]. All of these methods require precise and expensive equipment, with some of the methods only being able to be performed in the laboratory, and some requiring long test times [128]. This limits these methods to research activities and it excludes their use in routine runway surface texture assessments.

ICAO allows for the use of profilometry as a tool for macrotexture assessment [7]. During the profile assessment, the mean profile depth (MPD) is calculated, which can be converted to an estimated MTD in most cases. Profilometry estimated MTD usually differs from the volumetrically measured MTD because of the test methods, with volumetric methods affected by any deep pores in the material. For open graded courses, also known as porous friction courses, this difference can be significant and needs to be considered during the measurement [106]. In contrast, for dense graded asphalt and concrete, the two methods provide more comparable results [80].

There have been various attempts to develop a universal correlation between the volumetric and profilometry-based methods. However, different studies have reported different results. For example, during the PIARC experiment performed in Europe [104], a correlation between the two methods was developed. The correlation was used in different standards, such as [129,130]. However, subsequent studies reported that this correlation plot cannot be reliably used for the assessment of MTD [106]. For example, Figure 25 compares the PIARC correlation with measurements from independently tested surfaces. The difference is clear, particularly at MTD values above 2.0 mm.

In another study [80], the laser texture method was used to compare the conventional volumetric method of the MTD calculation to laser profilometer-based estimates of MTD. Figure 26 shows the correlation between MPD and MTD for all surfaces, which is clearly non-linear, as stated in the PIARC model. Furthermore, Figure 27 shows the improved correlation between a non-linear (quartic) conversion from MPD and an estimated MTD (referred by as ETD), which is consistent with the findings in Figure 25.

It is still a debatable question of which parameter is more representative, mean texture depth or mean texture profile, since these two methods represent two different properties of a surface, both of which are important for different elements of skid resistance assessment. The MPD estimates the braking force and drag available to the tire more precisely, because laser techniques are assessing the effective contact area between the pavement surface and the tire by omitting the effect of deep pores in the surface. In contrast, the MTD estimates the ability of the pavement to drain water from the pavement–tire interface, which is an important element of aircraft skid resistance in wet weather and is the focus of aircraft skid resistance management. Given that the two measurements focus on different surface characteristics, it is no surprise that a universal correlation between the two has not been found to be applicable to a broad range of surface types.

Full 3D imaging laser technology can also be used for macrotexture assessment as well. However, the area-based assessment, rather than a profile, allows for microtexture to be estimated as well as macrotexture if the precision is adequately high. For example, a low-precision scan was used to assess the macrotexture parameters of various surfaces [131]. In contrast, a high-precision laser was used to study both macrotexture and microtexture using a filtration algorithm in a different study [114]. The resolution of 3D measuring systems can reach 0.05 mm, which is comparable to the ICAO definition of microtexture [7].

Some researchers have proposed texture distribution as an alternate parameter, which can be defined as the ratio of the surface area and the plane area of the texture section [132]. Similarly, 3D scanning has been used to obtain the arithmetic mean curvature of the peaks and valley material portions for macrotexture and microtexture characterization [133]. Due to the high effectiveness of 3D scanning techniques, it is one of the most promising techniques in terms of friction assessment. The cost of this method, however, is higher compared to other measurement techniques [112] and it requires large and cumbersome equipment, making it less suited to field assessment.

6.2.3. Microtexture Assessment Using Non-Contact Methods

Microtexture assessment can be performed manually using a microscopic picture of the surface, with the shadow from a straightedge projected onto it at an angle. By measuring the profile of the shadow, it is possible to isolate and evaluate the microtexture manually. This approach has been used to assess the correlation between wear rate and microtexture of pavement surfaces [134,135]. The optical method with a straightedge shadow allows a precise microtexture assessment but is not suitable for large areas due to the manual evaluation, which requires time.

Microtexture assessment from photometry or 3D scanning can be challenging, due to the irregularity of the reference surface because of the macrotexture. Therefore, during the assessment, it is important to isolate the microtexture from the macrotextures. In a study performed by the Dutch Royal National Aerospace Centre [25], data obtained by a high-resolution surface texture scanner were used for the assessment of both microtexture and macrotexture. Consequently, researchers have developed algorithms for the separation of microtexture from the overall data. An example of separating the raw laser trace is shown in Figure 28. The surface laser scanner had a vertical resolution of 0.003 mm, a maximum length resolution of 0.00635 mm, and a maximum width resolution of 0.0247 mm. However, the drawback of this technique is the low scanning speed, due to the high resolution of the scanner.

At the same time, the microtexture assessment using the texture profile is overestimating the microtexture in the case of surfaces being worn [134]. Due to the wear of the texture in the contact area, microtexture depth can be significantly reduced. However, average texture depth can still show good results. For example, the average microtexture depth for every section of a profile was calculated for the data shown in Figure 28. The results (Figure 29) shown an average microtexture depth based on the microtexture profile. It can be seen that the lower microtexture depth occurred at the lowest point of macrotexture. Therefore, the average microtexture depth cannot provide an appropriate microtexture assessment. For this purpose, automatic or manual evaluation should be performed only in areas of contact between the tire and the pavement surface, where surface erosion has occurred. Otherwise, the result will be affected by the higher microtexture in the other areas.

In another study [136], white paint on the tire was used to evaluate the microtexture on the contact areas as well as the contact area itself. The results showed that the correlation coefficient between the British Pendulum Test results and the microtexture shape factor increases from 0.45 to 0.56 after restricting the measurements to only the contact parts of the tire footprint. In a similar study [137], the same problem was solved automatically by combining six different profiles from the same place. The resulting profile, which was referred to as the “summit profile,” was used for the friction assessment (Figure 30).

6.2.4. Other Methods

Another approach to macrotexture evaluation can be taken by measuring the surface drainage characteristics. For this approach, the outflow meter test can be used [138]. This does not measure the texture depth directly. Rather, it measures the ability of the depth and interconnected nature of the voids in the surface to let water pass through the surface of the pavement. The method was designed in the 1960s [40] and was widely used during research in the 1980s by NASA [139]. Nowadays, the outflow meter is rarely used, but it gives an assessment of drainage characteristics, which is good in terms of relative aquaplaning risk and prevention, reflected in the open-channeled nature of grooved surfaces, as well as the self-draining nature of porous friction courses, allowing aircraft operators to accept an aircraft braking performance credit in certain circumstances [140].

Surface texture can also be evaluated via mechanical effects, which can be a stylus, the chalk wear test, or a spot-friction assessment with a British Pendulum Tester. The stylus test is based on the mechanical measurement of the texture profile by registration of the stylus movement on a surface. This approach can be used for both macrotexture [141] and microtexture [142] assessments. It is widely used for different surface assessments, such as plastic or metal, but it can also be applied to pavement surfaces. The chalk wear test or similar tests are rarely used, but they can also be used for texture assessment [143]. This method is based on the measurement of chalk wear while sliding on the pavement surface. Deep and sharp texture, in this case, increases the chalk wear.

The British Pendulum Test is more common and can be used for microtexture assessment [144] due to the low speed of a pendulum. A recent study investigated the correlation between the ratings of the Global Reporting Format and the British Pendulum Test [145]. It was concluded that the British Pendulum Test correlated with the RWYCC value well. Another study, performed by the Federal Highway Administration, also showed a good correlation between the British Pendulum Number and microtexture shape factor divided by contact area fraction [136]. However, the reliability of the British Pendulum Test decreases with an increase in macrotexture [146]. This reflects the effect of the uneven surface on the pendulum swing. Consequently, it is not recommended to use the British Pendulum on surfaces with a high degree of macrotexture, such as sprayed seals [147] and stone mastic asphalt [148]. At the same time, reliable laboratory measurements with the closely packed aggregate required for the reliable test are likely to overestimate the skid resistance of surfaces because the aggregate spacing is wider than the spacing of the manually packed specimen used in the laboratory test [149].

6.3. Interaction Between Surface Texture, Runway Friction and Aircraft Skid Resistance

Surface texture plays a significant role in friction [150]. In runway design practice, most of the methods of friction improvement are based on texture improvement, which can be achieved by using a high macrotexture pavement surface layer or grooving [148,151,152]. This is because microtexture and the overall contribution of a runway surface to aircraft skid resistance cannot be directly measured or controlled. It is also well known that microtexture affects low-speed friction and macrotexture affects high-speed friction [153]. However, most aircraft skid resistance issues are associated with wet runways, with the dry surface having more friction and macrotexture than necessary. Therefore, macrotexture is a key focus for tire–surface contact area drainage.

In 1995, PIARC made an attempt to unify friction measurement practice by using texture parameters and the Penn State friction model [154]. This led to the development of the unified International Friction Index [104], as detailed in Equation (7).

$$F60=A+B\times FRS\times \exp \left[{\displaystyle \raisebox{1ex}{$-\left(60-S\right)$}\!\left/ \!\raisebox{-1ex}{$\left(a+b\times Tx\right)$}\right.}\right]+C\times T$$

(7)

where FRS is a friction testing result at speed S and Tx is a texture measurement result. A, B, C, and a, b are coefficients for particular friction and texture testing methods.

Device-specific coefficients were obtained with good correlation coefficients for each device [104]. Later, IFI calculation methods for new and different devices were standardized [130]. Despite the fact that some studies have found coefficients obtained by PIARC to be incorrect [105,106], the concept of IFI, as well as the dependence between friction testing results and texture measurements, have been proven to be valid [155,156]. A similar attempt was made by the Forum of European National Highway Research laboratories in 2006 as a further update of the International Friction Index method [157]. This method, however, has not become popular due to the same problems. The second problem of the International Friction Index and further similar studies is a lack of microtexture measurement techniques [73]. Besides the PIARC study, there have been various attempts to relate friction and texture parameters (

Table 8. Correlation between friction and texture parameters in different studies.

No.	Model	R2	Friction Assessment	Texture Assessment	Reference
1	$\Delta BPN=6.2 N_{B^0} {\left({\displaystyle \frac{Tp}{s}}\right)}^{-2.2}\times {10}^{-2}$	0.86	Difference between BPN value of flat surface and surface with macrotexture	Macrotexture depth and spacing between texture peaks	[158]
2	$SN=c_0\mathrm{e}\mathrm{x}\mathrm{p}\left(c_{1}v\right)$ $c_0=-31+1.38 BPN$ $c_1=-0.041 {MTD}^{-0.47}$	0.84	Skid Number according to ASTM E 274-70	Sand patch test and British Pendulum Test	[154]
3	$S{N}_{64}^R=0.766 BPN+4.72 MTD-9.7$ $S{N}_{64}^B=0.628 BPN+17.3 MTD-19.5$	0.92	Skid Number according to ASTM E 274-70 with ribbed and blank tire	Sand patch test and British Pendulum Test	[159]
4	$S{N}_{40}=18.6 TXT+12.6$ $TXT={\displaystyle \frac{S{F}_C\times 1000}{CA}}$ $s{F}_c={\displaystyle \frac{average peak height}{average peak width}}$	0.70	Skid Number according to ASTM E 274-70	Microtexture shape factor and contact area based on macrotexture	[136]
5	Rubber friction theoretical model [22] with modified data from optical measurement	0.91 and 0.97	ViaTech and Wehner/Schulze machine	Optical measuring system data	[21,137]
6	$\begin{array}{c}FN=42719 MP{D}^2-1508.8 MPD\\ +43.5\end{array}$	0.10 to 0.73	Locked-wheel trailer (speed higher than 45 mph)	Mean profile depth	[160]
7	Linear correlation models for different testing speeds and testing devices	0.56 to 0.79	Grip Tester	Mean profile depth	[161]
8	Artificial neural network model	0.88	Grip Tester	Laser profilometry (macro-textural parameters)	[162]
9	$\begin{array}{ll}{Y}_{Fr}={\beta }_0+{\beta }_{macro}& \times X_{macro}\\ & +{\beta }_{micro}\times X_{micro}\\ & +{\sum }_{i=i}^4{\beta }_i{X}_{type i}\end{array}$	0.43 to 0.82	British Pendulum Number, Grip Number, Dynamic Friction Test	Macrotexture and microtexture parameters, obtained by profilometry	[163]
10	$\mu =\phi +{\displaystyle \sum _1^8}{k}_i{\mu }_i$	0.78	Grip Tester	Texture parameters, obtained by 3D scanning data	[164]
11	Rubber friction theoretical model [22] with modified data from 3D scanning	0.60	British Pendulum Number	3D scanning data	[165]
12	$S_{MI}=BPN\times 0.0102+0.846$	0.84	British Pendulum Number	Microtexture index obtained by 3D scanning	[132]
13	$\begin{array}{ll}PT{V}_{predicted}=& -837.43+96.26 S_{dq,MIC}\\ & -852.14 S_{k,MIC}\\ & +10.41 S_{mr2,MIC}\\ & +4.92 S_{pc,MAC}\end{array}$	0.82	British Pendulum Number	Texture parameters, obtained by 3D scanning	[133]
14	Artificial neural network model	0.77 to 0.95	Dynamic Friction Tester	3D scanning data	[131]
15	Artificial neural network model	0.85	Sideway-Force Coefficient Routine Investigation Machine (SCRIM)	Sand patch test and 3D scanning	[114]
16	$\begin{array}{ll}BPN\left(AC\right)=9.235& +1899.789{ R}_{ami}\\ & +54.348 R_{ama}\end{array}$ $\begin{array}{ll}BPN\left(SMA\right)=& -91.573+10.938 CR\\ & +6547.885 R_{ami}\\ & +81.599{ R}_{ama}\end{array}$ $\begin{array}{ll}BPN\left(BBTM\right)=& -194.201+4.787 CR\\ & +2926.585 R_{ami}\\ & +296.675 R_{ama}\end{array}$	0.82 to 0.95	British Pendulum Number	Microtexture and macrotexture average roughness and rubber content for different mixes	[166]

).

Most of these studies appear to allow for the evaluation of friction using only texture parameters, without requiring friction to be measured at any reference or zero speed. Rather, to estimate the CFME value, various combinations of microtexture and macrotexture are used, with some studies using the BPN as an estimate of the microtexture contribution.

Other studies [21,137,165] use theoretical models for friction calculations. These studies are generally based on a hysteresis friction model [22]. This model also allows for the inclusion of contamination, such as water film thickness, using the same textural parameters, assuming that contamination acts as a filler for the texture, reducing the height of the asperities. These models also showed good correlation with CFME data. The Persson model can also be combined with numerical analysis to calculate the stresses in the contact area and increase the reliability of the model [167].

In some of the more recent studies [114,131,162], artificial neural networks were used for friction prediction. An artificial neural network is a good tool for friction analysis, and it shows a high correlation between predicted and measured values. However, one of the drawbacks of neural network models is that they work as a ‘black box’, since the calculation process is not presented for analysis. This creates a risk, since the result is based mostly on the training process and not on fundamental dependencies [168].

In addition to the studies presented in Table 8, there are also other studies of texture influence on friction, but they were not included due to their lack of numerical analysis of data or lack of experimental data. For example, one study investigated the texture skewness parameter [169], which was developed in early research [170,171]. This was found to be well correlated to the friction coefficient of the pavement. However, the calculation method for the friction estimation was not provided. In another study, the influence of texture on different parts of the runway on braking performance was studied [172]. It was concluded that macrotexture affects braking performance in touch-down zones more than microtexture. In a similar study [173], a correlation between macrotexture and measured friction was observed. However, the correlation was low due to lack of microtexture measurements. Finally, a correlation between texture parameters and friction was obtained based on the analysis of different parameters on sections of the Belgian road network [174]. However, the results contradict all other findings (Table 8), which cannot be explained by any theoretical model.

6.4. Summary

Current friction assessment methods are based on different techniques. The most significant methods for friction assessment are continuous friction measurement testing methods and volumetric methods for texture assessment. However, texture assessment methods are not fully developed in current international practice and some researchers have used various novel test methods, which are not yet considered in routine runway assessment.

Runway friction, at the same time, mostly depends on surface texture. Recent research analysis shows that texture assessment can reliably predict surface friction, including the effect of contaminants on aircraft skid resistance. These methods are based on theoretical models, phenomenological models, and artificial neural network models. These new techniques should be incorporated into routine runway characteristic measurement and management in the future.

Conventional friction assessment results, on the other hand, can be significantly affected by weather conditions and equipment and operator errors. With this in mind, it is recommended that an improved friction management system be developed, which includes modifications to texture assessment methods, friction reporting practices, and friction assessment, based on surface texture, which is discussed in the next section.

7. Improvement of the Runway Management System

The primary objective of this review was to determine improvements in runway friction, texture and aircraft skid resistance management for reduced aircraft skidding risk. Current friction assessment, runway maintenance, and surface condition reporting systems provide a reliable tool that can provide safety for aircraft and passengers. However, in this review it was found that the current ICAO friction assessment and skid resistance management system contains some elements that could be improved in the future. These include the following:

• There is no standard method or protocol for the determination of a characteristic value of macrotexture from multiple sand patch of other macrotexture measurements;
• There is no standard method or protocol for the determination of characteristic values of CFME wet friction from a CFME survey;
• The frequency of CFME surveys does not include a minimum frequency for regional airports with a low frequency of jet aircraft or airports with turboprop aircraft operations only, with many tested annually or when rubber contamination is visually identified;
• CFME-measured friction is significantly affected by temperature, meaning annual measurements are not necessarily representative of the annual cycle of friction;
• CFME measurements are not directly relatable to aircraft braking performance calculations because of the significant differences in vehicle speed, test wheel normal load and tire pressure of the CFME, compared to a typical aircraft;
• GRF does not allow adjustment of the RWYCC to intermediate friction levels;
• Microtexture is important, in combination with macrotexture, for aircraft skid resistance on wet runways, but microtexture is not routinely measured, due to historical equipment limitations, although new equipment and methods are now available, albeit some of these are time consuming and limited to laboratory use;
• Volumetric texture can be measured by contact methods (including deep pores and interconnected voids) and non-contact methods (excluding deep pores and interconnected voids) providing two important, but distinctly different, characteristics, making attempts to identify a universal conversion between the two impossible.

Some of these weaknesses could combine together or with other circumstances to lead to aircraft accidents, such as the Fokker F100 overrun at Newman Airport [54]. In this example, the runway friction levels were between the maintenance and minimum levels. Technically, that meant that no additional information about surface conditions was required to be reported to pilots. However, the actual surface friction of the wet runway surface was not provided. The historical friction survey results were likely affected by the annual variations in the measured friction levels as well, which led to an overestimation of aircraft braking capability. When combined with standing water in the wheel paths (Figure 7), and an extreme weather event, the runway overrun occurred. Another example is the Boeing 737-8FE overrun at Hobart airports, and this highlights the necessity of reliable and early microtexture assessment [175].

Potential elements of an improved friction assessment scheme have been developed and are presented in Figure 31, where arrows represent friction assessment and interpretation steps. In international practice, during the report of friction for landing and maintenance (µ_reported), two steps are presented. First, the actual friction level (µ₀) is assessed using CFME testing. This step is affected by weather conditions, equipment, and operator effects. This means that test results (µ_tested) contain an inevitable error. Next, the assessment results are interpreted in terms of maintenance requirements and landing restrictions. This step is also affected by errors in the interpretation of the results, due to imperfect estimation of the actual weather conditions, aircraft braking capability, or the pilot behavior, both the airport manager and the aircraft pilot. This means that errors are being accumulated and multiplied during the friction reporting (µ_reported).

To reduce these errors, it is proposed to also include texture assessment, instead of friction assessment. Texture assessment is not critically affected by errors because it is more reliable and repeatable. This assessment principle allows us to assess and report runway friction (µ_reported) without intermediate friction tests (µ_tested). Previous attempts to incorporate similar assessment systems were not successful due to the lack of reliable microtexture assessment tools [73]. Today, however, it is possible to solve that problem due to the development of various digital tools for microtexture assessment (Figure 24).

The proposed friction assessment methodology, however, requires modifications to the texture assessment methodology. It is important to establish a reliable macrotexture assessment methodology that allows for the use of modern texture assessment equipment. Microtexture assessment methodology also needs to be proposed. Finally, a texture measurement interpretation methodology should be provided. It is clear that wet friction levels are mostly affected by surface texture, and friction assessment methodologies based on texture assessment exist (Table 8). However, consensus on texture assessment results’ interpretations must be reached.

8. Conclusions

Aircraft skid resistance is a complex interaction of many factors and is important to aircraft safety. Because of the significant accidents that are related to skid resistance issues, this must be managed. Based on a review of regulations, there are mismatches between runway engineering regulation and assumptions in aircraft braking performance regulation. These need to be closed for consistency between the two elements of the aviation system. Runway regulations also contain contradictions, such as friction and macrotexture but not microtexture, high-friction surface options for aircraft credit, but no definitions of what these treatments need to achieve or how they are managed. These contradictions must be resolved.

International airport regulations rely primarily on CFME to achieve a minimum standard but there are issues with CFME such as climate, variability, and lack of specific interpretation protocols. Combinations of macro and micro are more reliable and should be fundamentally relatable to the braking available to aircraft in different weather conditions. But micro is not in the current regulations because such methods did not exist in the past. But microtexture measure is now possible and routinely available and this should be incorporated into the regulatory system.

In the future, improvement in the regulatory system is required by unifying and extending the friction measurement methodology. Two main goals need to be set: first, to increase the reliability of friction measurements, and second, to fundamentally relate friction measurements to aircraft braking performance. The reliability of friction measurements can be increased by considering such factors as the inconsistence of measurements on different parts of the runway, temperature, humidity, speed and loading during the measurement, and the influence of surface texture on friction measurement. In this review, examples of runway excursion incidents due to unreliable friction measurements have been identified and could have been avoided. The relationship between friction measurements and aircraft performance also needs to be established. The current GRF system allows pilots to be informed about the runway friction conditions, but that system does not include intermediate RWYCC values and consideration of CFME results. It was reported that, in some cases, the GRF system does not have the ability to prevent runway excursion accidents.

It is recommended that a new management approach be developed that includes surface texture measurements and calculation of a representative or characteristic friction value, based on those inherently variable measurements. Recent studies show that microtexture measurements can significantly increase the reliability of the friction assessment and this should be incorporated. The development of this approach, however, needs a modernization of current macrotexture assessment methods, and the development of a unified microtexture assessment tool. Moreover, it is obvious that improvement in the reliability of friction assessment increases the reliability of aircraft performance planning. Introducing these modern technologies can only increase aircraft and aviation safety by reducing the risk of aircraft overrun accidents in the future.