Last year (issue 157, November 2017), Rail Engineer contained an article with a very similar title covering tests using additional sanders and variable-rate discharge sand valves on two Class 387 EMUs at RIDC Melton. The article was upbeat, describing a multi-disciplinary, multi-company team of people working well together delivering useful tests with promising results.
This is the follow up – the “now the truth can be told” article – and the news is good. It’s sufficient to say that there was a strong requirement to understand sanding better and this series of tests was important. The results and challenges faced by the project team was the subject of a session in early February 2018 hosted by RSSB.
The objective was to establish whether the use of either more sanders and/or variable-rate delivery sanders could deliver or contribute to delivering a reliable brake rate of 6%g (0.6m/s2) in challenging adhesion conditions. If this rate could be achieved consistently, it would deliver the objective of predictable, seasonally agnostic braking, thus enabling year-round “services timed and delivered to the second”, and consistent and safe “running of trains closer together”. Both quoted statements are from the Rail Technical Strategy Capability Delivery Plan and are prerequisites of capacity improvement. There would also be safety and economic benefit.
The following tests were carried out (sander locations illustrated in figure 1):
- Establishing benchmark performance applying step-3 brake (9%g demand) where low adhesion conditions were created with a coefficient of friction (µ) down to 0.02 – a figure which, although extremely low, is encountered by most TOCs several times a day during the autumn;
- Performance of a standard Class 387 unit with axle-3 sanders – the standard configuration;
- Performance of eight-car train of two Class 387 units with both units’ axle-3 sanders enabled;
- Test of fixed rate sanders on axle 3 and on axle 7 or 11 on a four-car set;
- Test of variable rate sanders on axle 3 and axles 7 or 11 on a four-car set.
The fixed rate sanders dispense sand at the rate of 2kg/min with a cut off at 10mph. Variable-rate sanders dispense sand at the rate of 4kg/min down to 20mph and then ramp down to comply with the 7.5g/metre requirement, as illustrated in figure 2.
Both types comply with current standards, which were originally established to ensure train detection when operating over legacy low voltage DC track circuits.
The previous article described the testing process, which involved laying 96km of paper tape on the 1km low adhesion section, and running over 225 tests, of which 147 were sanded and 78 un-sanded. The brake cylinder pressures and axle speeds were monitored for all 16 (four-car) or 32 (eight-car) axles, as were train speed, doppler radar (for measuring speed independent of wheel speed), longitudinal acceleration and sander pressure. Video recordings were made of axle-3 sanders.
The results were presented in impressive detail and the highlights are summarised here.
Fig 3 shows how mean adhesion at each axle varies with each configuration of sanders against the base, no-sanding, low-adhesion condition. The result for three cases are described; no sand (black line), fixed-rate sanding on axles 3 and 7 (solid red line) and variable-rate sanding on axles 3 and 7 (dashed red line).
The average result for un-sanded tests shows µ for axles 1 and 2 of 0.03 (the same value for all configurations as axles 1 and 2 are not sanded) rising fairly smoothly to >0.04 at the back of the train. For the fixed-rate sanders, µ increases to 0.07 before falling back to 0.055 at axle 6.
The additional sand at axle 7 delivers µ of 0.09, which gently decays to a value of 0.065 for axle 16.
For variable-rate sanders at axles 3 and 7, µ increases to 0.085 at axle 3 and remains above 0.08 until axle 15.
The other cases not described in detail cover single axle-3 sanders (green lines – solid for fixed rate and dashed for variable rate) and sanders on axles 3 and 11 (blue lines – solid for fixed rate and dashed for variable rate).
Figures 4 and 5 (click to enlarge)
Figures 4 and 5 show the benefit of the increased adhesion in reduced stopping distance. The diagram shows stopping distance against initial reference adhesion for no sand (black diamond) single fixed-rate sander (green circle) and single variable-rate sander (green triangle). This shows that a useful benefit is delivered by changing single axle fixed-rate sanders to variable rate.
Dual variable-rate sanders (purple triangles) can deliver quite consistent stopping distances that are very close to the values that would be obtained on dry rail (horizontal dashes). The near-horizontal purple line was a very pleasant surprise to most people involved in adhesion management. They had been expecting a result closer to the green lines of figure 4.
All these results are for the 4-car Class 387 unit. Of course, many shorter trains run on the network, and the results for shorter trains were estimated. Assuming standard practice was to be adopted, that there will be at least six axles behind the last sander in order to manage the risk of failure to operate track circuits, likely configurations would be variable-rate sanders on axle 3 for a two-car unit and on axles 3 and 7 for a three-car unit. There was confidence that a three-car train would deliver results consistent with the four-car unit, but more work would be necessary to have the same confidence for the two-car unit.
Does this answer the exam question?
A resounding yes! For the four-car tests, with two variable rate sanders on axles 3 and 7, a deceleration rate of 6%g was consistently delivered in challenging adhesion conditions. This is clearly illustrated in figure 6, showing the incremental benefit of various sanding configurations when a step-3 full service brake is applied at 55mph with the initial reference adhesion of 0.02. Note, the un-sanded stopping distance of >1,200 metres is estimated as the low-adhesion test section was 1,000 metres long.
The project team concluded that dual variable-rate sanders could be implemented today for three-car, four-car and longer trains, whereas some further thinking is required for two-car units.
Would dual variable rate sanders use more sand? It was estimated that each axle’s variable-rate sanders would use no more sand than a single fixed-rate sander, so sand boxes would not need filling any more frequently, albeit there would be twice as many to fill.
The objective of demonstrating that 6%g braking can be consistently delivered has been achieved. For today’s railway, the findings suggest that SPADS and platform overruns caused by low adhesion could be reduced by over 90 per cent. For improving the railway of the future, this work provides important capabilities in order to move to a railway where services are timed to the second and trains run closer together.
Neil Ovenden of the Rail Delivery Group, and chair of the Adhesion Research Group, described this work as “the biggest advance in seasonal adhesion management in the last 20 years”, adding “we have finally quantified differences between, and benefit of, four different sanding configurations.”
The project can be declared a complete success and is a great tribute to the industry and all the hard work of everyone involved for pulling off such a complex programme. The next steps are to implement the results with pace and passion – a significant challenge in itself.
Implementing Dual Variable-Rate Sanders
The results of the dual and variable-rate sander tests are truly spectacular. However they will have no value unless implemented on the UK’s large passenger fleet. There will be in the order of 5000 individual units (2 – 11 cars) to modify, including a large number of new vehicles on order. This is a big challenge to the industry.
What has to be done? Based entirely on the author’s experience, at least the following tasks are required for each main class:
Business cases to be developed and agreed, together with securing funds where the medium to long term business case is good but where the short term case is poor (for example towards the end of a franchise);
Conclude the ideal configuration for two-car and units longer than four-cars (5,7,9,11);
Confirm that the ideal configuration for a four-car unit is appropriate for eight-car and 12-car trains composed of four-car units;
For some trains WSP (wheel-slip prevention) will need to be fitted;
Sanding and WSP suppliers will need to develop hardware (such as variable-rate valves) and software to control the sanders differently from today, followed by testing and validation;
System and application safety/assurance cases/technical files will need to be modified and assessed and, if the change is classed as significant, authorised by the regulator;
Change orders will need to be negotiated and contractors appointed;
Depots will need to be modified to handle more sanders and will probably need to store more sand.
Rail Engineer looks forward to reporting progress.
Thanks to Claire Grewer, Steve Mills and Paul Gray of RSSB, Liam Purcell – Ricardo Rail, Andrew Lightoller – DB ESG Rail, and Neil Ovenden – Rail Delivery Group.