Are we trying to produce a degraded ill-performing ACD in 30 days?
Dear Sirs, ( addressed to members in Railway Board Government of India)
To adopt the proven ACD system , in electrified multiple lines , only two issues are to be studies. How to fix the equipment in locomotives and guard vans, and how the locomotive electronics and ACD system interact without interference from each other. Some shielding and earthing issues will have to be addressed.
The reception of GPS under traction wires is already proven earlier in surveys conducted in these areas, but still that could be one item.
Then mounting the guard ACD, a portable unit, under wire has some electrical safety distance requirements to assure safety of a person placing the same. A non-conducting fibre glass rod mounted ACD with self locking in to position will suffice. It is not an insurmountable problem. This has to be demonstrated.
Then for multiple lines, the fixing of track ID some times could be a challenge purely based on deviation count, and so supplemental in formation through RFID tags and mounted readers was to be provided and this plug in is already provided in the ACD software developed earlier. This has to be reconfirmed.It is easy because it was already tested and proven while doing Skybus project.
So this trial should have been simple.
But shockingly I am told, that guard ACD is sought to be removed by advice from RDSO. That way some money can be saved. Further additional reliance on track circuits is being built in to the software for knowing train lengths etc because Guard ACD is to be eliminated. Then thirdly, repeater ACD towers in mid sections where, communication exchange distance between ACDs is affected by geography or alignment, not giving the adequate braking distance required to stop two trains , also is told to be removed. That means in mid sections, the ACDs will not be able to stop in time when such geographical features arise.
Further braking software and algorithm is being tinkered with. The ACD is not an automatic braking helper for the driver.
May be time for people to read the patent documents and description of the ACD network.
The implications I will summarise now:
1. Removal of Guard ACD.
This seriously compromises the very networking principle and definition of ACD, which is actually a network of 4 ACDs at any given time, when two trains are approaching or following each other. They are like sentinels at either end of train and in software this fact is used in various safety critical decisions. Cutting down 50% of the ACDs in the network reduces the ACD functionality and integrity very severely. I am surprised why my strong noting in files on this issue is over looked. This was considered and rejected at the highest levels. Rear end collisions and future up gradation to moving block systems, operational unpredictability of cutting and joining train units at some station, are various scenarios addressed by the provision of guard ACD. Actually there are 550 dimensions in the software simulation and without taking in to account the impact, across the table administratively taking such major decision is truly amazing. That means what is being tested is not the proven ACD in NF railway, but experiments for some new product of degraded functionality and performance.
2. Relying excessively on track circuits:
This is what is sought to be eliminated in ACD. As far as possible , the loco mouted equipment should be able to have capability to work out most of its information and should not depend on track based equipment. In unavoidable cases, limited use is made only to the extent of knowing track occupation. Even when track circuts fail ACDs act as safety shield in earlier design. Non-interlocked working cases will seriously affect ACDs too in this case and its presence is nullified. Latency in communication will also increase delaying decision making by the ACD. SO removing the guard ACD and building up reliance on track circuits is seriously compromising on the integrity of the ACD network.
3. Removal of repeater ACD:
At any given time the ACD communication should be such that adequate braking distance including time for response should be available for two running ACD fitted trains. After survey we locate such of the areas where repeater towers are located with ACD mounted to observe either direction , and if a ACD is approaching this information is stored and transmitted to another ACD approaching the vicinity, so that they analyse and decide if they are safely passing each other. WHen you remove this facility, then in such locations where communication is not upto mark, the integrity of ACD is compromised, because they will not stop as needed when an unfortunate incident ocurs. Coupled with removal of guard ACD , this problem will further accentuate, because the guard ACD dioes serve the purpose of repeater ACD too in a dynamic sense. The entire basis of ACD network's efficiency is sought to be removed by these steps.
4. Tinkering with braking algorithms.
The ACD has one of the most advanced braking algorithm in the world--self adjusting and corrects for the varying loads of trains and depends on the information from the guard ACD too. It is designed to come in to action when driver fails in normal course, where he should have acted. In such cases it has to be an emergency case, and so recorded . This will help point out the tendency of a driver to err and we can counsel them not to err. But if you make it too smooth and driver starts relying on the same willy nilly, we are violating a serious safety critical requirement, that driver shall not depend on non -signal equipment,
I provided only for two stage braking-- simple and effective. Smoothening efforts will result in too much variability because of the maintenance conditions in field for brake power are not exactly ideal. ACD does not replace the driver. It helps him and protects in danger.
This type of serious software changes should also undergo a set process of quality assurance and testing through ETDC. It is a a matter of 3 to 4 months.
If the above is true, I have no comment but to say it is truly less than intelligent innovation , who ever is trying it and too early to try.
That stage comes when ACD is certified for SIL and when moving block systems are planned. I foresee at least 5 years for this development to take place, provided we are serious. This childish pranks should not be tried when safety issues are involved.
My reckoning is , if the Board has agreed to these steps, they must understand what they are getting:
By seriously compromising on reliability and availability of inter-communication between ACDs, with the steps of removing guard ACD and repeater ACDs where needed, the ACDs are now incapable of giving assured protection to trains iin double line, and at gates too. The probability of failure is many orders increased. With only two ACDs in the network instead of four, the ACDs are compromised to lower level of performance.
That means ACDs are now reduced to watered down functioning only at stations and becomes a glorified TPWS type of equipment losing all its superior performance features.
So net result is Railway Board will still spend 70% of what it would cost with guard ACD and repeaters, for 30% of effectiveness.
More serious concern is after spending the amount serious accidents of collision are possible wherever communication is getting compromised thus, and then what answer the administration will have?
Such radical alterations are also being tried in a childish manner without first assuring the integrity of the new software developed for this purpose. As I know, the ETDC , a Government of India , software analysing and certifying authority, is associated , for the ACD clearance. But this is not done I am told.
Then full check of Lloyds in manufacturing and also for analysing results also is not being done. The reason is that there is no time.
Yes if one wants to do such unprofessional testing work, no product of reliability can be developed.
It is truly shocking that for such a serious issue, such careless approach is adopted and public money is being spent.
These trials will, naturally destined to, fail. If that is the objective, then trials are well designed.
I sincerely request that if you are serious about saving lives , then do it the right way. There is no short cut. Development of alternate software and to develop a new degraded ACD too is a matter of following a process correctly. It will take nothing less than 6 months.
But if you adopt proven ACD with limited objective of proving in electricfied areas and in multiple lines, it is simple. In the name of this trial, seeking to redefine and water down the ACD functionality is not only shocking but irresponsible in my considered opinion.
Best is do not waste money on such unreliable ACD , saving only 30% dropping the guard ACDs and repeater towers--- save full 100% not doing ACD. Because after fitting ACDs still you plan to leave open gaps by design, for collisions to take place. It is criminal.
May I request the CRB to kindly check up the facts and if true, please take necessary action. Let us not make fools of our selves in the international community trying to develop a degraded ill performing ACD in 30 days. This after getting a well performing one, after almost 7 years of hard work.
With regards,
--
सत्यं परम् धीमही
बी. राजाराम
Rajaram Bojji M.Tech. IRSE (Retd.), FIE, FNAE, AMASCE
Fmr. MD Konkan Railway /Min. of Railways India
www.atrilab.com
Comments