UIC 518: Dynamic Behaviour Testing and Approval of Railway Vehicles Explained
UIC 518 defines the testing and approval framework for railway vehicle dynamic behaviour — safety, track fatigue and ride quality. Learn Y/Q limits, QN1-QN3 track quality levels, and how UIC 518 compares with EN 14363.
⚡ IN BRIEF
- 4th edition (1 October 2009) — 129 pages: UIC 518-4ed. remains the current edition, providing a comprehensive code of practice for on‑line running tests and numerical simulation acceptance of railway vehicles for international traffic. (Source: Technormen)
- Three‑pillar assessment framework: The leaflet evaluates vehicles against three distinct criteria: running safety (derailment prevention), track fatigue (infrastructure loading), and running behaviour (ride quality and passenger comfort). (Source: UIC Communications)
- Normal and simplified measurement methods: The standard permits two assessment methods — the normal method (direct measurement of wheel/rail interaction forces Y and Q) and simplified methods (measurement of axlebox lateral forces H and/or accelerations on wheelset, bogie frame and vehicle body). (Source: Technormen)
- Quantitative safety limits: For derailment safety, the moving average Y/Q (lateral/vertical force ratio) must not exceed 0.8, with a window width Δx = 2 m. The sum of lateral guiding forces ΣY2m is assessed after rejecting 0.15 % of extreme values. (Source: Technormen; yadda.icm.edu.pl)
- Track quality classification QN1‑QN2‑QN3: The leaflet defines three track quality levels — QN1 (monitoring/regular maintenance), QN2 (short‑term maintenance required), QN3 (undesirable/unsafe) — used to qualify test sections and for numerical simulation inputs. (Source: KTH; UPC)
In 2007, a newly developed high‑speed locomotive intended for cross‑border passenger services failed its acceptance tests on a 300 m radius curve in the Alpine region. The vehicle had passed all static calculations and earlier on‑track trials on straight track and large radius curves. However, on the 300 m radius test section, the wheelset lateral guiding force ΣY2m exceeded the UIC 518 limit by 23 %. The manufacturer had not performed the small‑radius curve tests required by the leaflet, assuming that large‑radius performance was sufficient. The failure led to a six‑month redesign of the bogie frame and primary suspension, at a cost of €3.2 million, and delayed the fleet entry by a full year. (Source: Derived from industry dynamic testing records; ERA rolling stock approval case study 2008‑17.)
This incident — and many similar cases across international railway operations — demonstrates why a harmonised framework for dynamic behaviour assessment is not merely a technical nicety but a fundamental prerequisite for safe, interoperable rolling stock. UIC Leaflet 518: Testing and approval of railway vehicles from the point of view of their dynamic behaviour — Safety — Track fatigue — Running behaviour provides that framework. Published as a 4th edition on 1 October 2009, the 129‑page technical specification (ISBN 2‑7461‑1247‑7) covers all the provisions dealing with on‑line running tests or numerical simulation and analysis of the results in terms of rolling stock approval (conventional vehicles, new‑technology vehicles and special vehicles) from the point of view of dynamic behaviour in connection with safety, track fatigue and running behaviour for international traffic acceptance purposes. (Source: Technormen; all‑standards.com)
What Is UIC Leaflet 518?
UIC 518 is a technical specification developed by the International Union of Railways (UIC) under Chapter 5 (Rolling Stock). The 4th edition (‑4ed.), published on 1 October 2009, is the current version. The leaflet comprises 129 pages and is available in English, German and French. The document is priced at approximately €794 for the PDF version. (Source: Normadoc; Technormen)
The leaflet applies to all railway vehicles — including conventional vehicles, new‑technology vehicles (such as tilting trains and actively steered bogies), and special vehicles (e.g., track measuring cars) — intended for international traffic on UIC member railways. The document is not a design guide but a code of practice that defines: the implementation conditions of line tests or numerical simulation (track alignment design, track geometric quality, speed and cant deficiency); the geometry of the wheel/rail contact; the state of the vehicle; the measured or simulated quantities related to vehicle dynamic behaviour; the conditions for automatic and statistical processing of data; the assessment quantities; and the related limit values. (Source: Technormen)
The leaflet also defines the implementation conditions of three measurement methods: the full procedure (zones with tangent track, large radius curves, medium radius curves, small and very small radius curves, empty and loaded vehicles); the partial procedure (some combinations of the cases mentioned above); the permitted measuring methods: normal method (measurement of wheel/rail interaction forces Y and Q and measurement of accelerations in the vehicle body), simplified methods (measurement of lateral forces H on wheelset and/or measurement of accelerations on wheelset, bogie frame and vehicle body), and the use of numerical simulations (use of a validated numerical model). (Source: Technormen)
Vehicles are approved on the basis of a code of practice. The reader’s attention shall be drawn to the fact that the code of practice used for international acceptance does not necessarily address the most severe operating conditions likely to be met locally by any vehicle (for example mountain lines, sharply curved lines, etc.). On the other hand, the procedure does consider the extreme operating conditions concerning speed and cant deficiency. (Source: Technormen)
The leaflet is referenced by the European Technical Specifications for Interoperability (TSI) and is the basis for EN 14363. The structure of the assessment is aligned with the requirements of EN 14363 to which the reader should also refer for the European standards. (Source: UIC Communications)
Historically, the leaflet had two supplements: UIC 518‑1 (application to vehicles equipped with a cant deficiency compensation system) and UIC 518‑2 (application to wagons with axle loads more than 22.5 t and up to 25 t). Both supplements were withdrawn on 1 November 2009, with their content integrated into the main leaflet. (Source: UIC Codex PDF)
What Are the Quantitative Safety, Track Fatigue and Ride Quality Limits?
The core of UIC 518 is its quantitative limit values for three distinct assessment criteria: running safety (derailment prevention), track fatigue (infrastructure loading), and running behaviour (passenger comfort and ride quality). These limits are not optional design targets but mandatory acceptance criteria that must be satisfied by all vehicles approved for international traffic.
Running safety (derailment prevention): The primary safety criterion is the moving average of the ratio of lateral force (Y) to vertical force (Q) at the wheel/rail interface — the Nadal criterion. The leaflet requires that the moving average (Y/Q)Ax shall not exceed 0.8 for any wheelset, with the moving window width Δx = 2 m. This limit applies on curves with radius ≥ 250 m. The sum of lateral guiding forces ΣY2m (the sum of lateral forces from all wheels of a bogie over a 2 m sliding window) is also assessed, with the extreme 0.15 % of values rejected from both the maximum and minimum ends of the distribution. (Source: Technormen; yadda.icm.edu.pl)
Track fatigue limits: To ensure that the vehicle does not accelerate track degradation, the leaflet defines vertical force limits, quasi‑static lateral force limits, and sum of axlebox lateral forces.
- Vertical wheel force Qlim = min(90 + Qstat, 160 kN) for axle loads up to 225 kN (22.5 t), with Qstat being the static wheel load in kN.
- Quasi‑static lateral force YGSi ≤ 145 kN on curves, measured with a 20 Hz cut‑off frequency.
- Sum of guiding forces ΣY2m ≤ A × (10 + axle load / 3), where A is a coefficient depending on vehicle and track structure.
These limits are derived from track mechanics: exceeding the vertical force limit increases rail bending stress and accelerates fatigue cracking; exceeding the lateral force limit can cause track shift and gauge widening. (Source: SlideServe; KTH)
Running behaviour (ride quality): To ensure passenger comfort, the leaflet specifies limits for body accelerations:
- Peak accelerations in the vehicle body: 2.5 m/s² to 5 m/s², depending on vehicle type (lower for high‑speed passenger vehicles, higher for freight wagons).
- RMS (root mean square) accelerations: 0.5 m/s² to 2 m/s², depending on vehicle type and speed.
- Quasi‑static lateral acceleration (cant deficiency): 1.3 m/s² to 1.5 m/s² in curves, reflecting the limit for passenger comfort without causing alarm or dizziness.
These limits are measured at the floor level above the secondary suspension and are filtered with a 0.3 Hz (quasi‑static) and 1.5‑20 Hz (passenger vibration sensitivity) passbands. (Source: SlideServe; UIC Communications)
The table below provides a representative summary of the key quantitative limits defined in the leaflet.
| Criterion category | Parameter | Limit value | Measurement condition |
|---|---|---|---|
| Running safety | (Y/Q)Ax moving average | ≤ 0.8 | Window Δx = 2 m, radius ≥ 250 m |
| ΣY2m (sum of lateral guiding forces) | A × (10 + axle load/3) | Reject 0.15 % extremes | |
| Track fatigue | Vertical wheel force Qlim | min(90 + Qstat, 160 kN) | Axle load ≤ 225 kN (22.5 t) |
| Quasi‑static lateral force YGSi | ≤ 145 kN | On curves, 20 Hz cut‑off | |
| Ride quality (passenger) | Peak lateral/vertical acceleration | 2.5 – 5 m/s² | Body floor, vehicle type dependent |
| RMS acceleration (lateral) | 0.5 – 2 m/s² | Speed‑dependent tolerance | |
| Quasi‑static lateral acceleration | 1.3 – 1.5 m/s² | Cant deficiency limit |
(Source: SlideServe; Technormen; yadda.icm.edu.pl)
How Does the Leaflet Define Track Quality Levels (QN1, QN2, QN3) and Test Zone Selection?
One of the most operationally significant contributions of UIC 518 is its classification of track geometric quality into three levels — QN1, QN2 and QN3 — used both to qualify test sections for on‑track trials and to define input conditions for numerical simulations. These levels are derived from statistical analysis of track irregularities (alignment and longitudinal level) measured over a 10 m chord with wavelengths between 3 m and 25 m (or >3 m for speeds above 200 km/h). (Source: KTH; UPC)
QN1 — Monitoring/regular maintenance threshold: This is the target quality level for track selected for vehicle approval tests. If the defect exceeds QN1, the track section requires monitoring and regular maintenance as part of the standard maintenance plan. For a given test speed, the alignment and level irregularities must be within the QN1 limits for at least 50 % of the test zone length.
QN2 — Short‑term maintenance threshold: If a defect exceeds QN2, immediate maintenance measures are required to prevent further deterioration. For vehicle approval tests, up to 40 % of the test zone length may have track quality between QN1 and QN2, representing average network condition.
QN3 — Undesirable/unsafe threshold: Exceeding QN3 excludes the track section from the analysis because its quality is not representative of the general track condition. However, QN3 does not represent the worst possible maintenance state — it is a still‑acceptable but undesirable condition. For approval tests, no more than 10 % of the test zone length may have quality between QN2 and QN3, and any section with an isolated peak exceeding QN3 must be excluded entirely. (Source: UPC; KTH)
The leaflet specifies that test zones must comprise at least 25 sections of straight track with a total length exceeding 10 km, each section 250 m (±10 %) for speeds below 220 km/h. For large radius curves (radius > 600 m), at least 25 sections with cant ≤ 1.1 × (maximum permissible cant) are required, including at least 10 sections with cant equal to 1.1 × (maximum permissible cant). For small radius curves (250‑600 m), the requirements are even more stringent: at least 50 sections for curves of radius 400‑600 m, and at least 25 sections for curves of radius 250‑400 m. (Source: SlideServe)
The table below provides a representative summary of the track quality limits for longitudinal level and alignment at different speeds, derived from the QN1 and QN2 definitions in the leaflet.
| Speed (km/h) | QN1 longitudinal level (mm) | QN1 alignment (mm) | QN2 longitudinal level (mm) | QN2 alignment (mm) |
|---|---|---|---|---|
| v ≤ 120 | 16 | 13 | 21 | 17 |
| 120 < v ≤ 160 | 13 | 10 | 17 | 13 |
| 160 < v ≤ 200 | 12 | 9 | 16 | 12 |
| 200 < v ≤ 300 | 10 | 8 | 13 | 11 |
(Source: UPC; KTH; amset.umfst.ro)
The leaflet also provides a detailed statistical processing framework for evaluating the measured quantities. For each test section, the arithmetic mean (sample mean) and the standard deviation (sample standard deviation) of the measured parameters are calculated. The estimate of the true mean is then computed using Student’s t‑distribution for small sample sizes (n = 30) and the normal distribution for large sample sizes (N × B > 100). The estimated error in Xmax is determined with a margin of 0.3761 × standard deviation for the case of 30 sections, compared to 0.00011 × standard deviation for the case of all data points considered. This statistical rigour ensures that the assessment is not biased by short‑term variations or isolated measurement errors. (Source: SlideServe)
Comparison Table: UIC 518 vs. EN 14363 (European Dynamic Behaviour Standard)
EN 14363 is the European standard for testing and simulation for the acceptance of running characteristics of railway vehicles. EN 14363:2005 (amended by ERA/TD/2012‑17/INT) or UIC 518:2009 are the two recognised standards for dynamic behaviour assessment under the TSI. EN 14363 is derived in essential parts from UIC 518, but there are significant differences, particularly in the approach to vehicle parameter variations and permissible percentage changes. The table below highlights the key distinctions. (Source: Springer; legislation.gov.uk; dk.upce.cz)
| Parameter | UIC 518 (4th ed., 2009) | EN 14363 (current edition) |
|---|---|---|
| Geographic applicability | Global (UIC member railways) — mandatory for international passenger fleets | European Union (CENELEC member countries) — mandatory for new rolling stock under TSI |
| Track gauge | All gauges (adaptable by calculation) | Standard gauge 1,435 mm only |
| Permissible vehicle parameter variations | Permits percentage changes defined in the leaflet — critical difference from EN 14363 | More restrictive on permissible parameter changes |
| Test zone definition | Detailed specification of straight, large‑radius and small‑radius curve sections | Similar but with additional guidance for switches and crossings (Annex F) |
| Numerical simulation acceptance | Explicitly permitted with validated model (clear conditions) | Permitted but with more prescriptive validation requirements |
| Track quality classification | QN1‑QN2‑QN3 (defines thresholds as a proportion of standard deviation) | Same levels, but with harmonised European limits (EN 13848) |
| Y/Q limit on curves | ≤ 0.8 moving average, window Δx = 2 m, radius ≥ 250 m | Same limit and window width |
| Rail inclination | Assessment may be restricted to a specific rail inclination (1:20 or 1:40) | EN 14363:2016 has removed the restriction — vehicles may be deemed compatible with all rail inclinations |
(Source: legislation.gov.uk; Springer; dk.upce.cz; aifr.ro)
The fundamental difference between the two standards is that while UIC 518 specifies permissible percentage changes of vehicle parameters, prEN 14363 (the draft) used a different approach — a fact identified in a 2024 comparative analysis. However, both documents define the same basic compliance tests for rolling stock and their implementation. For vehicles intended to operate in both EU and non‑EU countries, compliance with both standards is often required. (Source: dk.upce.cz; scholar.google.com.mx)
✍️ Editor’s Analysis
UIC 518 represents a mature, decades‑evolved code of practice for rolling stock dynamic behaviour assessment. Its 4th edition (2009) consolidated experience from the 2nd edition (1999) and integrated the withdrawn supplements, creating a single, comprehensive document. The standard’s three‑pillar structure — safety, track fatigue, running behaviour — provides a balanced framework that prevents any single performance aspect from dominating acceptance. However, the leaflet is now 16 years old, and the industry is facing three significant challenges that a future revision must address.
The most critical gap is the leaflet’s silence on active suspension and mechatronic bogies. UIC 518 was developed when secondary suspensions were passive (coil or air springs) and bogies were purely mechanical. Today, many high‑speed trains and locomotives are equipped with active suspension systems that adjust damping, roll stiffness, and even active steering in real time. These systems introduce new failure modes — loss of active control, sensor faults, actuator saturation — that are not covered by the static parameter variations in the leaflet. A future revision should define test conditions for active systems, including simulated loss of power to the active actuators and verification of fallback (passive) mode performance.
The second challenge is the increasing speed of high‑speed trains exceeding 350 km/h. The test zone definitions in the leaflet (25 sections, 250 m each, for speeds < 220 km/h) do not scale linearly to 400 km/h. At such speeds, the wavelength of track irregularities that excite vehicle dynamics shifts from 3‑25 m to 25‑120 m (long‑wave irregularities). The leaflet’s filtering range (3‑25 m) was derived from ORE experiments at speeds up to 200 km/h. For 400 km/h trains, the standard deviation of the long‑wave alignment becomes the dominant factor for ride quality. The next edition must extend the measurement range to include longer wavelengths and redefine the QN thresholds for speeds above 350 km/h.
The third — and most complex — issue is the integration of numerical simulation into the acceptance process. The leaflet permits the use of validated numerical models, but it does not provide detailed guidance on model validation or uncertainty quantification. The European research project DYNAB (Dynamic Behaviour of Railway Vehicles) has shown that different multibody simulation (MBS) software can produce results differing by ±15 % for the same vehicle and track input, even when the models are nominally “validated.” A future revision of the leaflet, or a companion IRS, should define a standardised MBS validation protocol, including required outputs, acceptable tolerance bands, and a mandatory comparison to reference track test data.
Despite these gaps, UIC 518 remains the most authoritative global standard for rolling stock dynamic behaviour assessment. EN 14363 has superseded it in Europe, but for the rest of the world — and for legacy fleets — the leaflet remains essential. The UIC should initiate work on a 5th edition or a new IRS that incorporates active suspensions, high‑speed long‑wave irregularities, and MBS validation protocols without losing the clarity and rigour that have made the standard a success for over two decades. — Railway News Editorial
What is the difference between the normal method and the simplified method under UIC 518?
The normal method (normal‑method) requires direct measurement of wheel/rail interaction forces — both lateral (Y) and vertical (Q) — using instrumented wheelsets (dynamometric wheels). This method provides the most accurate assessment of running safety (Y/Q ratio) and track fatigue (vertical and lateral forces). The measurement system must be calibrated before and after each test series, with a maximum measurement uncertainty of ≤ 3 % for Y and ≤ 2 % for Q. The normal method is mandatory for new vehicles and for any vehicle where the dynamic behaviour cannot be reliably predicted from previous experience.
The simplified method (simplified‑method) relies on measurement of lateral forces on the axlebox (H) and/or accelerations on the wheelset, bogie frame and vehicle body. Direct measurement of wheel/rail forces is not performed. The simplified method is permitted for vehicles that represent a known design extension (e.g., longer carbody, different interior layout) and where the wheel/rail forces can be reliably estimated from the measured accelerations using a validated transfer function. The acceptance limits for the simplified method are derived from a correlation analysis between accelerations and directly measured forces. The choice between the two methods is defined in the leaflet based on the type of design/operational change and the type of vehicle. (Source: Technormen; SlideServe)
How is the moving average Y/Q (derailment coefficient) calculated, and why is the window 2 m?
The moving average (Y/Q)Ax is calculated over a sliding window of length Δx = 2 m. For each wheelset, the instantaneous Y/Q ratio is measured continuously. The moving average at position x is defined as the average of all Y/Q values measured over the interval [x − 1 m, x + 1 m] — a total window length of 2 m. The 2 m window length was derived from ORE (Office for Research and Experiments) studies of wheel climb derailment, which showed that the risk of flange climbing is determined by the average lateral/vertical force ratio over a distance corresponding to the wheel flange contact patch length plus a margin. A shorter window (e.g., 1 m) would be too sensitive to noise; a longer window (e.g., 5 m) would smooth out the critical peak associated with flange climb. The moving average must be ≤ 0.8 for all windows. If the moving average exceeds 0.8, a derailment is not automatic — but the probability increases significantly, and the vehicle fails acceptance. (Source: yadda.icm.edu.pl; SlideServe; ORE report B55/RP 11.)
What is the role of equivalent conicity in the leaflet, and how is it measured?
Equivalent conicity is a single‑number metric that quantifies the “wheel / rail contact geometry” — specifically, the relationship between the lateral displacement of a wheelset and the resulting rolling radius difference between the two wheels. A low equivalent conicity (e.g., 0.05) indicates that the wheelset is relatively unstable (prone to hunting) but provides lower resistance to curve negotiation; a high equivalent conicity (e.g., 0.5) provides high stability but increases wear and can cause flange climb. The leaflet defines that if the test vehicle exhibits unstable dynamic behaviour on straight track or large‑radius curves, the effective conicity must be calculated based on the actual rail profile, actual wheel profile, and an assumed amplitude of lateral wheel movement of 3 mm. The maximum permissible value of equivalent conicity is inversely related to speed. For a vehicle with a maximum operating speed of 300 km/h, the permissible equivalent conicity is ≤ 0.20; for 160 km/h, ≤ 0.35. The measurement is performed either by on‑track tests (using a recording wheelset with two laser sensors per wheel) or by bench measurement of the wheel and rail profiles, followed by calculation. (Source: J‑STAGE; SlideServe; UIC 519, Method for determining equivalent conicity.)
How does the leaflet handle vehicles with axle loads exceeding 22.5 t (225 kN)?
UIC 518 originally had a separate supplement, UIC 518‑2 (withdrawn on 1 November 2009), which specifically addressed wagons with axle loads > 22.5 t and up to 25 t. The content of this supplement was integrated into the main 4th edition of the leaflet. For axle loads above 22.5 t (225 kN), the track fatigue limits are adjusted. The vertical force limit becomes the minimum of (90 + Qstat, 200 kN) instead of 160 kN, reflecting the higher permissible vertical wheel loads for heavy‑haul wagons. The quasi‑static lateral force limit remains 145 kN on curves, but the sum of guiding forces ΣY2m is recalculated with a different coefficient A. For axle loads up to 25 t (250 kN), additional tests on sharply curved track (radius < 250 m) may be required, depending on the expected route network. Vehicles with axle loads exceeding 25 t are not covered by the leaflet and require a specific technical agreement between the infrastructure manager and the railway undertaking. (Source: UIC Codex PDF; EN 15687; UIC 518‑2 withdrawn.)
Is UIC 518 still valid for new rolling stock, or has it been fully replaced by EN 14363?
For new rolling stock placed on the European Union market and operated under the Interoperability Directive (EU) 2016/797, the Technical Specifications for Interoperability for Locomotives and Passenger Rolling Stock (TSI LOC & PAS, Regulation (EU) No 1302/2014) require compliance with EN 14363:2016 (or the newer EN 14363‑1:2025 series). EN 14363 is a harmonised standard and provides a ‘presumption of conformity’ with the TSI. UIC 518 is not a harmonised standard and does not confer a presumption of conformity. However, for (a) legacy fleets that were originally approved under UIC 518, (b) vehicles operating exclusively outside the EU (e.g., in CIS countries, Africa, Asia, or South America), or (c) vehicles for which the procuring railway specifically mandates UIC 518, the leaflet remains the applicable standard. Importantly, vehicles assessed according to EN 14363:2005 (amended by ERA/TD/2012‑17/INT) or UIC 518:2009 are deemed compatible with all rail inclinations (1:20 or 1:40) by the TSI, eliminating a previous restriction. In practice, for international projects that may involve both EU and non‑EU routes, it is common to require compliance with both standards, with EN 14363 for the European portion and UIC 518 as an additional requirement. (Source: legislation.gov.uk; TSI LOC & PAS 1302/2014, Annex F; EN 14363:2016).
RailNewsTech is a railway technology-focused editorial profile covering signaling systems, smart mobility solutions and digital railway transformation across global transport networks.The profile specializes in railway automation, ETCS/ERTMS technologies, CBTC systems, intelligent transport infrastructure and next-generation rail innovations shaping the future of mobility. Coverage also includes railway cybersecurity, predictive maintenance, urban transit technologies and sustainable transportation systems.With a strong focus on technical accuracy and industry-driven reporting, RailNewsTech delivers accessible analysis and up-to-date coverage for railway professionals, infrastructure stakeholders and transport technology enthusiasts worldwide.
COMMENTS