Real-World Exhaust Emissions of Diesel Locomotives and Motorized Railcars during Scheduled Passenger Train Runs on Czech Railroads

Round 1

Reviewer 1 Report

This paper is on an interesting topic, since information about locomotive emissions under real world operating conditions is still limited. The paper provides some good insight into their operation and different considerations related to locomotive testing. An important consideration of this paper is that the authors are discussing the possibilities of the utilized measurement system for broader testing. Since the system is not based on the PEMS systems used for regulatory testing such as RDE or NTE testing, additional details on the performance of the analyzers compared to laboratory tests should be provided, expanding upon the text that is already provided on that topic. Other comments are discussed below along with edits in a attached PDF file.

 

1. Introduction –

 

There is a single sentence paragraph in line 96 that needs to be placed elsewhere where is can be a completed thought/idea.

 

Throughout the paper, the terms “car” and “vehicle” are used somewhat often. This can be confusing as these terms are also associated with smaller passenger cars and vehicles. Can the cars by renamed railcars, or another appropriate name, and “vehicles” be renamed locomotives.

 

2. Experimental –

 

Many of these suggestions are related to better defining the accuracy of the measurement system.

 

Section 2.3 – 3^rd paragraph – It talks about speed-density method being reasonably accurate based on refs [32,34]. Can one more sentence be added as to how accurate the method was in terms of percent differences.

 

Section 2.3 – 4th paragraph – It talks about the analyzer being “comparable” to laboratory measurements. Another sentence or two could be added here given the percentage differences, so that the accuracy of the measurements in the paper and being proposed more broadly can be understood.

 

Section 2.3 – 4th paragraph – Relating to the light scattering instrument. It discusses the instrument is “somewhat proportional” to PM mass but not necessarily at a unity slope. For the calibration that was done for the light scattering during steady state operations, what was the slope of the calibration curve? Can this be included in the text. Given the typical elemental/organic nature of locomotive smoke, would the light scattering be more of less affected relative to other sources of PM. It later talks about nephelometer measurements 65% higher than the gravimetric measurements.

 

Section 2.3 – 5th paragraph – Again, the authors talk about providing “somewhat quantitative” values for mean particle diameter and total PM.

 

Section 2.3 – 6th paragraph – Presumably the authors got the lug curve for locomotive 854.

 

3. Results –

 

4^th paragraph – just below figure 11 in line 401. It talks about using extrapolation of computed regression lines to infer stabilized values. Another sentence or two could be added to better clarify how this was done.

 

Paragraph above Figure 14. Can the authors add some discussion of the upward bend of the Nox curve at higher fuel use values.

 

Paragraph above table 4. It talks about the emissions per kilometer per passenger, but those values are not included in the table.

 

4. Discussion –

 

3rd paragraph – It talks about how the distribution of engine loads resembles those of the prescribed tests. It would be interesting to see how the in-use distribution of loads differs from that of the prescribed cycles. How would the weighting factors be changed if the weighting factors are based on in-use distributions. Another comparison Table of these distributions would be interesting.

 

Last paragraph in section 4.1 – The paragraph starts with a sentence that the FTIR and NonMet did not have any substantial benefits, but the rest of the paragraph is a bit more positive for these instruments. Probably a reworded first topic sentence would be more appropriate.

 

First paragraph in section 4.2 – There a phrase “in complex terrain frequently lower, in this study.” That does not make sense, and should be reworded.

 

2nd paragraph in section 4.2 – There is a sentence “Another possibility could be the combustion timing.” Is there evidence that one is retarded/advanced relative to the other?

 

Section 4.4 – The second sentence talks about a per mass comparison, but of what?

 

5. Conclusions –

 

3^rd paragraph – For the emissions on a per passenger-km basis. Perhaps this should be clarified that this is based on automobiles. It would be useful to include that they are comparable to Euro 2-5 for NOx and to diesel cars without DPFs for PM.

 

6. edits –

 

A separate PDF is provided with edits.

Comments for author File: Comments.pdf

Author Response

We thank the reviewer for very detailed comments and for correcting the English.

To clarify, railroad cars are referred to as „railcars“, highway passenger cars are referred to as „automobiles“, and rail cars, units and locomotives, when discussed as a group, are referred to as „rail vehicles“.

The single sentence at line 96 has been deleted.

All comments pertaining to accuracy/validation of measurement:

The Experimental section has been expanded to elaborate on the validation of the monitoring system. The traditional tests – comparison of one measured compound at steady-state conditions – are necessary but not sufficient, as such conditions are not particularly relevant. Instead, measured emissions are evaluated over a dynamic cycle, where both the concentrations and the exhaust flow vary considerably, and an agreement between the instruments typically necessitates simultaneous fulfillment of accurate measurement of both concentrations and exhaust flow, adequate time response, and synchronization of multiple data streams. Surrogate measurements of PM are discussed.

Section 2.3 – 6th paragraph – Presumably the authors got the lug curve for locomotive 854.

We obtained the engine brake-specific fuel consumption map for the Caterpillar 3412 E-DITA engine as a part of manufacturer’s documentation.

4^th paragraph – just below figure 11 in line 401. It talks about using extrapolation of computed regression lines to infer stabilized values. Another sentence or two could be added to better clarify how this was done.

Expanded text now reads: For three selected longer uninterrupted accelerations at notch 8, the concentrations of particulate matter and NO_x are shown in Figure 11. It is apparent that in none of the cases a steady-state value is reached. All type approval and in-use tests use, however, steady-state values obtained at stabilized operation at a given notch. For comparison with legislative tests, stabilized values can be reasonably inferred by extrapolation of computed regression lines, shown in Fig. 11 along with regression equations obtained by an iterative process discussed in [45]. 

Paragraph above Figure 14. Can the authors add some discussion of the upward bend of the Nox curve at higher fuel use values.

The simultaneous relative increase in NO_x and a relative decrease in PM resembles a typical shift along the diesel engine NO_x-PM curve, and therefore appears to be a result of engine calibration, not necessarily applicable to rail engines in general.

Added text: The simultaneous relative increase in NO_x and a relative decrease in PM resembles a typical shift along the diesel engine NO_x-PM curve, and therefore appears to be a result of engine calibration, not necessarily applicable to rail engines in general.

Paragraph above table 4. It talks about the emissions per kilometer per passenger, but those values are not included in the table.

It has been clarified that emissions per pkm are not included in the table.

3rd paragraph – It talks about how the distribution of engine loads resembles those of the prescribed tests. It would be interesting to see how the in-use distribution of loads differs from that of the prescribed cycles. How would the weighting factors be changed if the weighting factors are based on in-use distributions. Another comparison Table of these distributions would be interesting.

Discussion added into the text, comparing with US EPA line haul duty cycle: On the Prague to Tanvald route, approximately 46% of time was spent at idle and approximately 18% at full load (notch 8), compared to 50.5% at idle and 16.2% at notch 8 specified for line-haul locomotive tests in the U.S. legislation [47].

Last paragraph in section 4.1 – The paragraph starts with a sentence that the FTIR and NonMet did not have any substantial benefits, but the rest of the paragraph is a bit more positive for these instruments. Probably a reworded first topic sentence would be more appropriate.

Reworded: The use of FTIR and NanoMet is believed to be of key significance for new technologies and fuels, and was tested here despite higher power consumption and mass to the more simple monitoring system, which was capable of providing comparable NO readings. 

First paragraph in section 4.2 – There a phrase “in complex terrain frequently lower, in this study.” That does not make sense, and should be reworded.

Reworded: ... compared to the track speeds of 80 or 100 km/h, and even lower in river gorges and other areas with poor visibility, in this study.

2nd paragraph in section 4.2 – There is a sentence “Another possibility could be the combustion timing.” Is there evidence that one is retarded/advanced relative to the other?

There is no such evidence, but in general, higher NOx and lower PM, all other things being equal, are a typical sign of advanced combustion timing. An explanation has been added to the text.

Section 4.4 – The second sentence talks about a per mass comparison, but of what?

Mass of the vehicle per passenger. Clarified in the text.

3^rd paragraph – For the emissions on a per passenger-km basis. Perhaps this should be clarified that this is based on automobiles. It would be useful to include that they are comparable to Euro 2-5 for NOx and to diesel cars without DPFs for PM.

Clarified. The text reads: Despite all engines approaching the end of their life, NO_x and PM emissions per passenger-km were very surprisingly low compared to those from European diesel automobiles (Euro 2-5 without a particle filter, but also the Czech “fleet average”).

The language comments have been corrected as suggested.

 

Reviewer 2 Report

The study by Vojtisek-Lom et al. is based on measurements of exhaust emissions of particulate matter, nitrogen oxides and other pollutants from two diesel-electric locomotives. Measurements were performed with self-made but evaluated monitoring instruments, which perform well in the lab and in real-world. The measurements are plausible although they measure very low PM emissions. Nevertheless, the reason for this outcome is discussed in detail.

In total, the study is an admirable piece of work, with a manuscript written in good english (for a non-native speaker), an easy-to-follow structure with a lot of details, very good figures and valuable outcomes for the emissions and air quality modeling community. Besides this, the application of a self-designed can be considered as another valuable outcome for future measurements.

Although it becomes clear, that the study was conducted with a lot of effort and the manuscript tries to reflect this, the manuscript is somewhat long and is sometimes more a good story than a scientific publication. Thus, I recommend to shorten and dense the manuscript to scientific valuable information and maybe move some very detailed descriptions on machines and processes to the supplement.

I do fully support this study and recommend publication after minor revisions, which are listed below.

 

Line 38: How do you evaluate the density based on network length? 9567 km lines per x? Per Population or per area? Comparison to other countries or an an average would strenghten this argument in the first sentence.

line 217: please add manufacturer and type of instruments (analyzer and scattering instrument). 

Table 1 and table 2 are really a valuable output for the emissions and air quality modeling community. Gratulations again.

line 553-556: can you cite literature for this?

 

We would like to acknowledge again, this very well conducted study with valuable results and contribution to the scientific community.

Author Response

Line 38: How do you evaluate the density based on network length? 9567 km lines per x? Per Population or per area? Comparison to other countries or an an average would strenghten this argument in the first sentence.

Text expanded: The Czech Republic, with 9567 km of active railroad lines [1], has one of the highest density railroad infrastructures in the world, both on per area (over 12 km per 100 km²) and per capita (over 9 km per 10 000 inhabitants) basis.

line 217: please add manufacturer and type of instruments (analyzer and scattering instrument). 

Details added, the text reads:

Emissions were measured with a home-made monitoring system constructed by the first author, utilizing components typical for a BAR 97 standard (Bureau of Automotive Repair, www.bar.ca.gov) garage-grade five-gas analyzer and a laser light scattering instrument (modified model 8587A, TSI, St. Paul, MN, USA), sampling raw, undiluted exhaust (...)

The gas analyzer employed a non-dispersive infra red (NDIR) cell (modified version of Andros 6500, Lumasence, CA, USA, approved to BAR 97 standards) to measure the concentrations of hydrocarbons (HC. 0-10 000 ppmC), carbon monoxide (CO, 0-10%) and carbon dioxide (CO₂, 0-16%), and an electrochemical cell to measure nitric oxide (NO, 0-5000 ppm).

line 553-556: can you cite literature for this?

References added: Given the large difference in test cycle and real-world operating conditions, the test cycles may also be prone to “cycle beating“, a questionable but in recent history not uncommon practice of carelessly or even deliberately ”tuning“ the engine control unit so that the emissions are higher during real-world operation than during the prescribed test cycles [32,48-53].

 

Reviewer 3 Report

Repetition in line 249 Line 598: "observer" should be "observed" In line 483 the Authors state that "For 749 and 754, electric heating increased, on the average, the fuel consumption by 22 kg/h, and NOx emissions by 1.5-2.1 kg/h." Was this measured separately in fully controlled conditions (at standstill, idle, with only heating turned on with no other auxiliary systems active)? Or was it inferred from the general measured data obtained during measurements? This can greatly affect the uncertainties involved in the values provided. The emissions measuring apparatus was described as "home-made" and therefore naturally has unique measuring properties. This means that the obtained data, while extremely useful in the discussed aspects, cannot be directly related or compared to similar measurement results performed with different measuring systems. Information regarding the device measuring range and emission data accuracy obtained with it would enable a wider comparison. Such information, if determined and if it can be disclosed, would greatly increase the scientific value and international interest of the article.

Author Response

Repetition in line 249 Line 598: "observer" should be "observed"

Corrected.

In line 483 the Authors state that "For 749 and 754, electric heating increased, on the average, the fuel consumption by 22 kg/h, and NOx emissions by 1.5-2.1 kg/h." Was this measured separately in fully controlled conditions (at standstill, idle, with only heating turned on with no other auxiliary systems active)? Or was it inferred from the general measured data obtained during measurements? This can greatly affect the uncertainties involved in the values provided.

This was inferred from the general data; the consumption of the other auxiliaries was much smaller, intermitted, beyond our control, and assumed to be independent of heater operation.

Text expanded: For 749 and 754, electric heating increased, on the average, the fuel consumption increased by 22 kg/h, and NO_x emissions increased by 1.5-2.1 kg/h (depending on the engine temperature), calculated as the difference between the idle power consumption during periods with heating and during periods near the end of the run when the heating was not active. Much smaller variations are attributable to automatic operation of the air compressor and of the auxiliary cooling fans. 

The emissions measuring apparatus was described as "home-made" and therefore naturally has unique measuring properties. This means that the obtained data, while extremely useful in the discussed aspects, cannot be directly related or compared to similar measurement results performed with different measuring systems. Information regarding the device measuring range and emission data accuracy obtained with it would enable a wider comparison. Such information, if determined and if it can be disclosed, would greatly increase the scientific value and international interest of the article.

The Experimental section has been extended. The following has been added:

Analogous home-made and commercially produced monitoring systems have been used for on-road studies over the last two decades and have undergone extensive comparison testing. As a part of instrument validation for a California roadside truck study [35], total NO_x and CO₂ were measured by the portable system (using calculated exhaust flow) and by a laboratory (using a full-flow dilution tunnel) on a light duty diesel truck over multiple transient cycles driven multiple times. The correlation of total emissions per test (for the three monitoring systems, slopes were 1.05-1.06 for NO_x and 0.94-0.96 for CO₂, Pearson’s R² coefficients were 0.991-0.997 for NO_x and 0.990-0.998 for CO₂). A similar comparison using three full-size diesel pickup trucks and a greater range of test cycles has shown a greater variance [34], approximately up to 15-20% for NOx, and, in most cases, approximately 20-30% for PM (quantitative data not given). As the total mass emissions per cycle are influenced by the concentration measurements, exhaust flow inference, and synchronization between concentration and exhaust flow data, a good agreement between the instruments on mass emissions over a transient cycle was taken as a validation of concentration and exhaust flow computations.

The monitoring system used here was tested at the the state certification laboratory TUV-SUD Auto in Prague on a Euro 3 Iveco Tector highway diesel engine during both steady-state and transient tests. The intake air mass flow and the concentrations of CO, CO₂ and NO_x were comparable with the laboratory measurements (correlation of second-by-second data over European Transient Cycle: intake air flow: slope 0.986, R² = 0.961, CO₂: slope 0.964, R² = 0.956, NO_x: slope 0.949, R² = 0.73, CO: slope 0.932, R² = 0.627), except for the slower response for NO_x, especially during decreasing concentrations, and for CO at very low concentrations (below about 0.02%).

The monitoring system has also undergone extensive comparison testing at the departmental engine laboratory on a Zetor 1505 tractor engine with a mechanically controlled inline injection pump and no aftertreatment, certified to approximately 4 g/kWh NO_x and 0.3 g/kWh PM. The laser scattering method, when used with raw, undiluted exhaust, pumped at a relatively fast rate, reheated to prevent condensation of water in the instrument, provided a reading that was, during comparison tests, mostly within 20-25_% of PM mass [36] when running on diesel fuel.