HSTEd
Veteran Member
- Joined
- 14 Jul 2011
- Messages
- 16,810
Well, for a long time now I have been talking about how advances in the design and performance of power electronics would effect the economics of third rail electrification and now it is time for me to finally write up one of my proposals.
This is actually a rather simpler proposal than I originally had in mind because I realised that the alternative to providing a fancy booster electronics to draw return current out of the rail and thus increase the effective supply voltage was rather unnecessary when we can simply provide more substations.
The normal cricitism of third rail technologies is that they normally require substations at relatively close spacings of only a few kilometres to keep supply currents and voltage drops within reason.
This tends to increase the capital cost of the installation relative to spacings of 50km or more that can be obtained in 25kV systems.
There are two approaches to try and keep the voltages under control - one is to attempt to reduce the loop impedance of the circuit by paralleling cables with the running rails (far more effective than doing so with the conductor rail as the running rails have a far higher starting impedance).
The second approach is to simply move the substations closer together.
Traditional 1MW+ substations are simple voltage source systems that require HV type supplies in the rectifier hut with all the relevant grounding and other infrastructure that comes with that. Indeed Network Rail operates a large network of 33kV feeder cables specifically to provide supply to its traction system.
Portland Streetcar in the United States pioneered the concept of massing small substations with a power of ~300kW that could be supplied by the low voltage network. This drastically reduced the size and complexity of the substations - and effectively their cost.
This approach is the one that I intend to apply in this example.
For the purposes of compiling estimates I have assumed that the specimen train to be used is a Class 319 with a maximum traction load of approximately 1000kW delivered. These trains are available in substantial quantity and would likely be the workhorse of any near-term third rail extension - should one be authorised.
The third rail standards document used in this example is a copy of Railway Group Standard GM/RT1001 Issue 1 'Classic 750V d.c third rail Electrification sSystem and T&RS parameters to ensure Interworking'. Although this document dates from the mid-1995 and has obviously been withdrawn it was the copy I have to hand and is thus instructive in outline, if not in detail.
As this standard dates to after the introduction of the Class 319 it seems highly likely that the train will function properly if the supply is kept within the parameters laid out within it.
Example Line - modelling assumptions
The system is modeled as twin-track plain line without junctions, however the running rails and conductor rails are all paralled together respectively at intervals of 100m or so, allowing the cross bond resistance to be neglected and both conductor rails and running rails to be modelled as entirely shorted together. This is estimated based on the Delta Rail report DeltaRail*ES*2010*003 - 'Low Cost Electrification for Branch Lines' to cost roughly £5k/single track km. Or £10km/route-km.
The system utilises high conductivity 18kg/m aluminium-stainless steel conductor rails and twin-rail traction current return, the cost of instalaltion of which varies between ~£50k/track-km (in the case of utilising existing sleepers modified for conductor rails and insulated chairs) and ~£500k/track-km (in the case of total resleepering and ballast renewal).
That takes our track preparation costs to somewhere in vicinity of £110k-£1m/route-km depending on if the existing sleepers and ground conductivity conditions produce acceptable stray-current performance. It is clear that the size of this cost in the specific case to be considered will be the making or breaking of the business case.
All trains on the route are considered to consist of a single Class 319 unit with 1000kW of installed traction power. This represents a significant increase in comparison with two or three vehicle diesels, many with shorter vehicles, that are often utilised on many of the exemplar routes in the North West of England and elsewhere.
Due to the lightly trafficked nature of the route regeneration is assumed to be disabled, as permitted by the Group Standard above.
As all conductor and running rails are paralleled at regular intervals the overral route loop-resistance between a substation and train-load can be modelled as approximately 13 milli-ohm/km, as described in the Railway Group Standard referenced above.
Substations are placed at 1000m intervals and consist of 300kW DC-power variable-voltage substations, but it is instructive to describe the operating mode of this substation design here.
It will operate in a constant voltage mode at an output voltage of approximately 850V until total substation DC-power reaches approximately 300kW, and this time the output voltage of the substation will drop as the current increases such that the substation does not overload and continues to output 300kW.
Worst case load model
As the trains consist of 1000kW-max point loads the worst case scenario that can be envisaged is that of two units accelerating away from a single station in opposite directions. As this is only likely to occur in a station, which would be the likely location of a substation, it will be assumed that the 2000kW total load is positioned at a substation.
It is further assumed, for the purpose of iteration, that the voltage at the pick up shoes of the units is 825Vdc [it is much simpler to calculate this and determine if your substation voltage is sufficient than the reverse].
In order to provide 2000kW at 825Vdc it is necessary for the traction systems on the two units to require a total of 2424A.
As the local substation is not anywhere near capable of producing this the substation output voltage will sag in a controlled manner [to 825V in this example] and the substation will produce 364A. This will cause approximately 2060A to be drawn to the units through the conductor/return rails from the directions of the adjacent substations.
Assuming the substations are evenly spaced and have similar loop impedances 1030A will be drawn from both directions.
[This case is also useful for determining the end-of-line condition with a single set, as a train is highly unlikely to be accelerating at full power into the buffer stops!]
It is 1000m to the next station, generating a loop impedance to the next station of approximately 13 milli-ohm.
That means that 13.4V will drop between the central substation and the first one working outwards in each direction [as the system is symmetric]. This means the voltage seen by this substation is 839Vdc
As the substation can only provide 358A at that voltage the remaining 672A will flow from further along the conductor rail.
It is 1000m to the second substation out, at 13 milli-ohm impedance a further 8.74V will drop. This means the second substation sees a rail voltage of 848Vdc. At this voltage level the substation is capable of providing 354A - as such the remaining 318A will flow along the conductor rail from the third and final substation.
The last 13 milli-ohm 1000m section sees a voltage drop of 4.13V drop, taking the voltage drop at the final rectifier to ~853Vdc, with the substation being capable of providing the entire remaining current requirement. THis requires the substation to produce roughly 272kW of DC-power.
853Vdc is well within the rating of the RGS standard, indeed an 855Vdc substation voltage would still provide substantial opportunity for regeneration if it was enabled on route as the maximum permanent voltage in the standard is 900Vdc.
In this worst-case situation there are 7 substations involved in supplying the 2000kW of traction power to the two units in question. 5 operate at their full rated 300kW power whilst the remaining two produce 272kW. That translates to a total DC power of 2044kW.
~600V chopper rectifiers in industry have been demonstrated to have efficiencies of at least 97.5% - which will be covered later.
This means that something approaching ~2100kW of AC power is required to produce the 2000kW of traction power. An overall efficiency of ~95% in this very severe scenario. THis is comparable to AC systems and demonstrates the advantages of providing current to a train from close by - even if the entire train supply cannot be handled in such a manner.
Provision of electricity supplies
Clearly this approach is radically different to that which is normally pursued in the creation of such substations.
Because of the physically greater number of substations required for this technique, one per route-km, it is necessary to reduce the cost of each one significantly.
One way of reducing costs, as mentioned previously, is to take a supply from the electricity utility at Low Voltage. This is beneficial as an electricity utility normally has lower costs associated with maintenance of HV equipment and transformers due to not requiring rail qualified staff and due to economies of scale that come with operating virtually all the substations in a given area.
The examples from the Electricity Northwest document 'Statement of Methodology and Connection Charges (5 December 2016)' were used to formulate approximate values for the cost of providing multiple ~350kVA LV supplies at positions scattered along the railway. These are not exact by any means but merely attempt to provide a conservative estimate of the costs. AS these improvements are required to supply the customer (the railway) with a minimum service, the railway would be required to pay for the required works, either by the utility or by another authorised contractor.
AS the railway will pass through rural areas in most cases providing a HV supply from which to draw a supply may prove challenging, therefore it is assumed that an 11kV HV feeder has been provided parallel to the railway, as this is unlikely to be required along the entire route it is assumed that this value includes necessary off-route connections. An estimate drawn from information in the document suggests that a cost of roughly £100k/km would be suitable for an overhead line system running adjacent to the railway.
The provision of 350kVA LV compact substations and metering equipment, as well as terminals for the short LV service cable to the substation can be estimated based on the stated values at approximately ~£40k per example.
This would suggest that costs relating to the provision of an electricity supply is roughly ~£140k/km.
As the Point of Connection for each substation would be at 400V the railway would not be liable for network reinforcement costs above the 11kV level, as stated in the methodology provided by Electricity North West.
Substation design
The substations would be of a chopper-rectifier design with an input 12-pulse low voltage transformer.
These systems function as transformer-isolated diode rectifiers with what amounts to a buck-converter on the output of the diode rectifier. This allows them to maintain an excellent input power factor whilst providing a highly controllable output, voltage and current of the supply output can be monitored at all times and the operation of the chopper-converter can be controlled to keep them within the required values.
They are capable of operating at at least 97.6% in the 600Vdc+ range utilising existing silicon IGBTs and it is highly likely that higher efficiencies could be obtained by utilising newly available Silicon Carbide diodes and MOSFETs in the output stage (1700V SiC MOSFETs are available and are suitable for this application).
Additionally the higher junction temperature tolerable with silicon carbide devices and their inherently higher thermal conductivity will permit more compact substations to be designed which will further tend to reduce costs.
Example costs for a 300kW-output rectifier, based on the experience of Portland Streetcar and in the opinion of Delta-Rail, seems likely to be in the range £300k per example. That translates to £300k/route-km. However it should be noted that as numerous identical unit substations would be acquired under any reasonably sized scheme it is highly likely that costs could be reduced drastically.
Additionally due to the low input voltage it seems highly likely that significant economies could be made by adopting a medium-frequency transformer topology with the attendant reduction in the costs of substation magnetics.
Other Costs
Delta-Rail's low cost electrification branch study suggested that other costs could become important, such as the cost of signalling immunisation and the provision of a SCADA system to allow the control of the electrification scheme and the like. It was suggested that immunisation of signalling would cost roughly ~£20k/route-km, but this is obviously heavily dependent upon the signalling system in use at the time. It also seems likely that the eventual arrival of ERTMS-Regional would permit the route to operate without the track circuits that can make signalling systems on newly electrified routes so difficult to maintain and efficienctly operate.
Additionally a price of £100k/route-km was suggested for a SCADA cable along the route that could connect all the substations to a control centre to enable the system to be operated effectively and to enable inter-tripping and other such techniques to be used to improve safety. However with the arrival of the safety grade GSM-R network on the Railway it is possible that this cable could be deleted in favour of simply providing an inexpensive GSM-R data base station at each substation. However this option would be subject to further development and it seems likely the first scheme would utilise a conventional control system.
This suggests other track-related costs to be between £0-£120k/km depending on the conditions prevailing on the route at the beginning of the project.
It is also suggested that ~£2-3m be allocated for project costs, largely indepedently of the size of the scheme proposed (within reason obviously!).
Summary of costs:
Track and conductor rail costs: £110-£1010k / route-km
Provision of electricity supply: £140k / route-km
Substation costs: £300k / route-km
Other costs: £0-120k / route-km + £2-3m total.
Overall costs: £2-3m + (£550-1570k / route km.)
Compared to something like £4m/route-km for a 25kV scheme at the present time, based on examples of schemes being turned out today.
It should be noted that even in the worst case listed above the trains have an enormous amount of margin before they violate the terms of the third rail standard, and should they remain more than ~7-8km apart they should not interact with each other through the traction system.
This means that the system can support a twin track route with several four-car trains per hour in each direction without much difficulty. Even conservatively with 10km average spacings between following trains a line with a ~40km/hr average speed could maintain 4 trains per hour without any driver constraints at all.
This makes it easily capable of handling all likely traffic loads on many secondary routes near existing northern third rail territory.
It is also startling how much difference the existing track condition makes - although it should be made clear that total resleepering would delay the next required resleepering operation in the future - which reduces the railway's future costs and as such some portion of the cost of that operation should be defrayed by the normal railway operational budget as opposed to the electrification budget.
It also seems likely that any future resleepering operations on routes near third rail territory could be a golden opportunity to electrify the routes in question, taking advantage of advances in the technologies mentioned above.
Just my two cents - which comes from reading far too much about power electronics in industry in recent years. And a love for messing around with a pad of paper, a calculator and an engineering problem.
This is actually a rather simpler proposal than I originally had in mind because I realised that the alternative to providing a fancy booster electronics to draw return current out of the rail and thus increase the effective supply voltage was rather unnecessary when we can simply provide more substations.
The normal cricitism of third rail technologies is that they normally require substations at relatively close spacings of only a few kilometres to keep supply currents and voltage drops within reason.
This tends to increase the capital cost of the installation relative to spacings of 50km or more that can be obtained in 25kV systems.
There are two approaches to try and keep the voltages under control - one is to attempt to reduce the loop impedance of the circuit by paralleling cables with the running rails (far more effective than doing so with the conductor rail as the running rails have a far higher starting impedance).
The second approach is to simply move the substations closer together.
Traditional 1MW+ substations are simple voltage source systems that require HV type supplies in the rectifier hut with all the relevant grounding and other infrastructure that comes with that. Indeed Network Rail operates a large network of 33kV feeder cables specifically to provide supply to its traction system.
Portland Streetcar in the United States pioneered the concept of massing small substations with a power of ~300kW that could be supplied by the low voltage network. This drastically reduced the size and complexity of the substations - and effectively their cost.
This approach is the one that I intend to apply in this example.
For the purposes of compiling estimates I have assumed that the specimen train to be used is a Class 319 with a maximum traction load of approximately 1000kW delivered. These trains are available in substantial quantity and would likely be the workhorse of any near-term third rail extension - should one be authorised.
The third rail standards document used in this example is a copy of Railway Group Standard GM/RT1001 Issue 1 'Classic 750V d.c third rail Electrification sSystem and T&RS parameters to ensure Interworking'. Although this document dates from the mid-1995 and has obviously been withdrawn it was the copy I have to hand and is thus instructive in outline, if not in detail.
As this standard dates to after the introduction of the Class 319 it seems highly likely that the train will function properly if the supply is kept within the parameters laid out within it.
Example Line - modelling assumptions
The system is modeled as twin-track plain line without junctions, however the running rails and conductor rails are all paralled together respectively at intervals of 100m or so, allowing the cross bond resistance to be neglected and both conductor rails and running rails to be modelled as entirely shorted together. This is estimated based on the Delta Rail report DeltaRail*ES*2010*003 - 'Low Cost Electrification for Branch Lines' to cost roughly £5k/single track km. Or £10km/route-km.
The system utilises high conductivity 18kg/m aluminium-stainless steel conductor rails and twin-rail traction current return, the cost of instalaltion of which varies between ~£50k/track-km (in the case of utilising existing sleepers modified for conductor rails and insulated chairs) and ~£500k/track-km (in the case of total resleepering and ballast renewal).
That takes our track preparation costs to somewhere in vicinity of £110k-£1m/route-km depending on if the existing sleepers and ground conductivity conditions produce acceptable stray-current performance. It is clear that the size of this cost in the specific case to be considered will be the making or breaking of the business case.
All trains on the route are considered to consist of a single Class 319 unit with 1000kW of installed traction power. This represents a significant increase in comparison with two or three vehicle diesels, many with shorter vehicles, that are often utilised on many of the exemplar routes in the North West of England and elsewhere.
Due to the lightly trafficked nature of the route regeneration is assumed to be disabled, as permitted by the Group Standard above.
As all conductor and running rails are paralleled at regular intervals the overral route loop-resistance between a substation and train-load can be modelled as approximately 13 milli-ohm/km, as described in the Railway Group Standard referenced above.
Substations are placed at 1000m intervals and consist of 300kW DC-power variable-voltage substations, but it is instructive to describe the operating mode of this substation design here.
It will operate in a constant voltage mode at an output voltage of approximately 850V until total substation DC-power reaches approximately 300kW, and this time the output voltage of the substation will drop as the current increases such that the substation does not overload and continues to output 300kW.
Worst case load model
As the trains consist of 1000kW-max point loads the worst case scenario that can be envisaged is that of two units accelerating away from a single station in opposite directions. As this is only likely to occur in a station, which would be the likely location of a substation, it will be assumed that the 2000kW total load is positioned at a substation.
It is further assumed, for the purpose of iteration, that the voltage at the pick up shoes of the units is 825Vdc [it is much simpler to calculate this and determine if your substation voltage is sufficient than the reverse].
In order to provide 2000kW at 825Vdc it is necessary for the traction systems on the two units to require a total of 2424A.
As the local substation is not anywhere near capable of producing this the substation output voltage will sag in a controlled manner [to 825V in this example] and the substation will produce 364A. This will cause approximately 2060A to be drawn to the units through the conductor/return rails from the directions of the adjacent substations.
Assuming the substations are evenly spaced and have similar loop impedances 1030A will be drawn from both directions.
[This case is also useful for determining the end-of-line condition with a single set, as a train is highly unlikely to be accelerating at full power into the buffer stops!]
It is 1000m to the next station, generating a loop impedance to the next station of approximately 13 milli-ohm.
That means that 13.4V will drop between the central substation and the first one working outwards in each direction [as the system is symmetric]. This means the voltage seen by this substation is 839Vdc
As the substation can only provide 358A at that voltage the remaining 672A will flow from further along the conductor rail.
It is 1000m to the second substation out, at 13 milli-ohm impedance a further 8.74V will drop. This means the second substation sees a rail voltage of 848Vdc. At this voltage level the substation is capable of providing 354A - as such the remaining 318A will flow along the conductor rail from the third and final substation.
The last 13 milli-ohm 1000m section sees a voltage drop of 4.13V drop, taking the voltage drop at the final rectifier to ~853Vdc, with the substation being capable of providing the entire remaining current requirement. THis requires the substation to produce roughly 272kW of DC-power.
853Vdc is well within the rating of the RGS standard, indeed an 855Vdc substation voltage would still provide substantial opportunity for regeneration if it was enabled on route as the maximum permanent voltage in the standard is 900Vdc.
In this worst-case situation there are 7 substations involved in supplying the 2000kW of traction power to the two units in question. 5 operate at their full rated 300kW power whilst the remaining two produce 272kW. That translates to a total DC power of 2044kW.
~600V chopper rectifiers in industry have been demonstrated to have efficiencies of at least 97.5% - which will be covered later.
This means that something approaching ~2100kW of AC power is required to produce the 2000kW of traction power. An overall efficiency of ~95% in this very severe scenario. THis is comparable to AC systems and demonstrates the advantages of providing current to a train from close by - even if the entire train supply cannot be handled in such a manner.
Provision of electricity supplies
Clearly this approach is radically different to that which is normally pursued in the creation of such substations.
Because of the physically greater number of substations required for this technique, one per route-km, it is necessary to reduce the cost of each one significantly.
One way of reducing costs, as mentioned previously, is to take a supply from the electricity utility at Low Voltage. This is beneficial as an electricity utility normally has lower costs associated with maintenance of HV equipment and transformers due to not requiring rail qualified staff and due to economies of scale that come with operating virtually all the substations in a given area.
The examples from the Electricity Northwest document 'Statement of Methodology and Connection Charges (5 December 2016)' were used to formulate approximate values for the cost of providing multiple ~350kVA LV supplies at positions scattered along the railway. These are not exact by any means but merely attempt to provide a conservative estimate of the costs. AS these improvements are required to supply the customer (the railway) with a minimum service, the railway would be required to pay for the required works, either by the utility or by another authorised contractor.
AS the railway will pass through rural areas in most cases providing a HV supply from which to draw a supply may prove challenging, therefore it is assumed that an 11kV HV feeder has been provided parallel to the railway, as this is unlikely to be required along the entire route it is assumed that this value includes necessary off-route connections. An estimate drawn from information in the document suggests that a cost of roughly £100k/km would be suitable for an overhead line system running adjacent to the railway.
The provision of 350kVA LV compact substations and metering equipment, as well as terminals for the short LV service cable to the substation can be estimated based on the stated values at approximately ~£40k per example.
This would suggest that costs relating to the provision of an electricity supply is roughly ~£140k/km.
As the Point of Connection for each substation would be at 400V the railway would not be liable for network reinforcement costs above the 11kV level, as stated in the methodology provided by Electricity North West.
Substation design
The substations would be of a chopper-rectifier design with an input 12-pulse low voltage transformer.
These systems function as transformer-isolated diode rectifiers with what amounts to a buck-converter on the output of the diode rectifier. This allows them to maintain an excellent input power factor whilst providing a highly controllable output, voltage and current of the supply output can be monitored at all times and the operation of the chopper-converter can be controlled to keep them within the required values.
They are capable of operating at at least 97.6% in the 600Vdc+ range utilising existing silicon IGBTs and it is highly likely that higher efficiencies could be obtained by utilising newly available Silicon Carbide diodes and MOSFETs in the output stage (1700V SiC MOSFETs are available and are suitable for this application).
Additionally the higher junction temperature tolerable with silicon carbide devices and their inherently higher thermal conductivity will permit more compact substations to be designed which will further tend to reduce costs.
Example costs for a 300kW-output rectifier, based on the experience of Portland Streetcar and in the opinion of Delta-Rail, seems likely to be in the range £300k per example. That translates to £300k/route-km. However it should be noted that as numerous identical unit substations would be acquired under any reasonably sized scheme it is highly likely that costs could be reduced drastically.
Additionally due to the low input voltage it seems highly likely that significant economies could be made by adopting a medium-frequency transformer topology with the attendant reduction in the costs of substation magnetics.
Other Costs
Delta-Rail's low cost electrification branch study suggested that other costs could become important, such as the cost of signalling immunisation and the provision of a SCADA system to allow the control of the electrification scheme and the like. It was suggested that immunisation of signalling would cost roughly ~£20k/route-km, but this is obviously heavily dependent upon the signalling system in use at the time. It also seems likely that the eventual arrival of ERTMS-Regional would permit the route to operate without the track circuits that can make signalling systems on newly electrified routes so difficult to maintain and efficienctly operate.
Additionally a price of £100k/route-km was suggested for a SCADA cable along the route that could connect all the substations to a control centre to enable the system to be operated effectively and to enable inter-tripping and other such techniques to be used to improve safety. However with the arrival of the safety grade GSM-R network on the Railway it is possible that this cable could be deleted in favour of simply providing an inexpensive GSM-R data base station at each substation. However this option would be subject to further development and it seems likely the first scheme would utilise a conventional control system.
This suggests other track-related costs to be between £0-£120k/km depending on the conditions prevailing on the route at the beginning of the project.
It is also suggested that ~£2-3m be allocated for project costs, largely indepedently of the size of the scheme proposed (within reason obviously!).
Summary of costs:
Track and conductor rail costs: £110-£1010k / route-km
Provision of electricity supply: £140k / route-km
Substation costs: £300k / route-km
Other costs: £0-120k / route-km + £2-3m total.
Overall costs: £2-3m + (£550-1570k / route km.)
Compared to something like £4m/route-km for a 25kV scheme at the present time, based on examples of schemes being turned out today.
It should be noted that even in the worst case listed above the trains have an enormous amount of margin before they violate the terms of the third rail standard, and should they remain more than ~7-8km apart they should not interact with each other through the traction system.
This means that the system can support a twin track route with several four-car trains per hour in each direction without much difficulty. Even conservatively with 10km average spacings between following trains a line with a ~40km/hr average speed could maintain 4 trains per hour without any driver constraints at all.
This makes it easily capable of handling all likely traffic loads on many secondary routes near existing northern third rail territory.
It is also startling how much difference the existing track condition makes - although it should be made clear that total resleepering would delay the next required resleepering operation in the future - which reduces the railway's future costs and as such some portion of the cost of that operation should be defrayed by the normal railway operational budget as opposed to the electrification budget.
It also seems likely that any future resleepering operations on routes near third rail territory could be a golden opportunity to electrify the routes in question, taking advantage of advances in the technologies mentioned above.
Just my two cents - which comes from reading far too much about power electronics in industry in recent years. And a love for messing around with a pad of paper, a calculator and an engineering problem.
Last edited: