Vehicles
JONAS ERIKSSON SIMON FAGERHOLM
KTH Industrial Engineering and Management
Master of Science Thesis Stockholm, Sweden 2014
Exhaust Analyser for Simplified Emissions Testing on Heavy Duty Vehicles
Jonas Eriksson Simon Fagerholm
June 19, 2014
Master of Science Thesis MMK 2014:50 MDA 476 KTH Industrial Engineering and Management
Machine Design
SE-100 44 STOCKHOLM
Heavy Duty Vehicles Jonas Eriksson Simon Fagerholm
Approved: Examiner: Supervisor:
19-06-2014 Lei Feng Mikael Hellgren
Commissioner: Contact person:
Scania CV AB Markus Walseth
.
Abstract
.Over the years the regulations on emissions from heavy duty vehicles have become stricter.
Emission measurements during development are therefore done by the manufacturers in order to check compliance in an early stage. This is a time demanding process due to complicated installation and operation of the test equipment. Therefore a simplified concept for replac- ing such equipment was evaluated and a prototype was designed capable of measuring NOx
concentration, exhaust volume flow and CO2 concentration together with reading on board diagnostic messages over CAN in order to include additional truck sensory data. This re- port describes the development process and contains modular tests of the prototypes separate modules as well as a whole systems test against Horiba OBS-2200, a portable emissions mea- surement system, in a truck. The results from the tests were that NOxconcentration, exhaust volume flow, CO2 concentration and NOxaccumulated mass showed good correlation with the references used. The prototype system achieved a coefficient of determination, R2 value, of 0.988 for the NOx mass flow compared to the reference and a total error of less than 7% of accumulated NOx mass.
fordon Jonas Eriksson Simon Fagerholm
Godkänt: Examinator: Handledare:
19-06-2014 Lei Feng Mikael Hellgren
Uppdragsgivare: Kontaktperson:
Scania CV AB Markus Walseth
.
Sammanfattning
.Med introduktionen av Euro VI har kraven på lägre utsläpp från tunga fordon skärpts. För att kunna uppnå dessa krav är det naturligt att emissionstester utförs även på utvecklingsfordon som en del av utvecklingsprocessen. Dessa tester är tidskrävande på grund av komplicerad installation och svårhanterad utrustning. För att minska problemen med utvecklingstester utvärderades ett nytt koncept för mätning av NOx koncentration, avgasflöde och CO2 kon- centration samt avläsning av OBD-data. Denna rapport beskriver utvecklingsprocessen av en prototyp och utvärdering av moduler samt test av hela prototypen mot Horiba OBS-2200, en emissionsmätutrustning som används på Scania CV AB, på en lastbil. Resultaten visar att NOxkoncentration, CO2 koncentration, avgasflöde och ackumulerat NOxmassutsläpp kan mätas med bra överensstämmelse mot referensutrustningen. Prototypen uppnådde en deter- minationskoefficient på 0.988 för massflöde av NOx relativt referensen och hade ett totalt fel på mindre än 7% i ackumulerad NOx-massa.
with them. It has been a real pleasure! Secondly to thank our company supervisor Markus Walseth for his invaluable help and patience throughout these past 20 weeks.
From KTH we would like to thank our supervisor Mikael Hellgren for his help with this project.
A special thanks should go to Hans Svensson for his help with our software and Nils Lindberg and Sven Andersson for their insight into the testing process. Other people at Scania we would like to thank are Göran Rundkvist for lending us the NOx testing equipment, Petra Larsson and Tony Karlsson for letting us use the acoustic flow rig. Also the whole NME department deserves recognition for their support and interest in our project.
Lastly we would like to thank Jonas Nordh for his feedback on our designs and for manufac- turing our special tail pipe.
Jonas Eriksson and Simon Fagerholm Södertälje, June 2014
Contents
Abstract i
Sammanfattning iii
Foreword v
Contents vii
List of Figures ix
List of Tables xi
Nomenclature xiii
1 Introduction 1
1.1 Background . . . 1
1.2 Problem Description . . . 1
1.3 Purpose & Goals . . . 1
1.4 Limitations . . . 2
1.5 Methodology . . . 2
1.5.1 Project Structure . . . 2
1.5.2 Rejected Alternative Methods . . . 3
2 Frame of Reference 5 2.1 Interview . . . 5
2.1.1 Design . . . 5
2.1.2 Compilation of Answers . . . 5
2.2 Horiba OBS-2200 . . . 6
2.2.1 Performance . . . 7
2.3 Gas Flow Measurement . . . 8
2.3.1 Problem Description . . . 8
2.3.2 Solutions . . . 9
2.4 NOx Measurement . . . 16
2.4.1 Problem Description . . . 16
2.4.2 Solutions . . . 16
2.5 CO2 Measurement . . . 22
2.5.1 Problem Description . . . 22
2.5.2 Solutions . . . 22
3 Concept 25 3.1 Comparison and Evaluation . . . 25
3.1.1 Flow Measurement . . . 25
3.1.2 NOx Measurement . . . 26
3.1.3 CO2 Measurement . . . 27
3.2 Final Design . . . 27
3.3 Flow Accuracy Calculations . . . 29
4 Implementation 31 4.1 Functional System Overview . . . 31
4.1.1 Main Unit . . . 31
4.1.2 Software . . . 32
4.2 Tail Pipe Attachment . . . 32
4.3 Sensors . . . 33
4.3.1 Differential Pressure Transducer . . . 33
4.3.2 Absolute Pressure Sensor . . . 34
4.3.3 Temperature Sensor . . . 34
4.3.4 Zirconia NOxSensor . . . 34
4.4 Verification . . . 35
4.4.1 NOx Modular Test . . . 35
4.4.2 Gas Flow Modular Test . . . 37
4.4.3 Whole System Performance . . . 39
5 Results 41 5.1 NOx Modular Test . . . 41
5.1.1 Uncalibrated Results . . . 41
5.1.2 Uncalibrated (Fused) . . . 43
5.1.3 Calibrated Results . . . 45
5.1.4 Calibrated (Fused) . . . 48
5.2 Gas Flow Modular Test . . . 51
5.2.1 PEMSLight Flow Measurement Technique Compared to Horiba . . . 51
5.2.2 Swema Positioning and Influence . . . 52
5.3 Whole System Performance In Truck . . . 52
5.3.1 Signal Correlation . . . 52
5.3.2 Gas Flow . . . 55
5.3.3 CO2 Estimation . . . 61
5.3.4 NOx . . . 62
6 Discussion & Recommendation 67 6.1 Discussion . . . 67
6.1.1 Prototype Criteria Evaluation . . . 67
6.1.2 Differences in Flow Measurements . . . 67
6.1.3 NOx Deviations Compared to the Horiba . . . 67
6.1.4 NOx Fusion Strategy and Linear Calibration . . . 68
6.1.5 Flow Estimation Using λ . . . 68
6.1.6 Novel Flow Estimation . . . 69
6.2 Future Work . . . 70
7 Conclusion 71
8 Bibliography 73
Appendices 77
A Data sheets 79
List of Figures
2.1 Example of mounting the Horiba system in a truck . . . 7
2.2 Simplified overview of the Horiba system . . . 7
2.3 Cross-section of a pipe showing flow direction and dimensions . . . 9
2.4 Figures of different pitot tubes . . . 10
2.5 Geometrical overview of the transit-time flow meter . . . 13
2.6 Laminar flow in pipe section . . . 14
2.7 Overview of the principle behind a chemiluminescence system . . . 17
2.8 Nernst cell principle showing the transfer of oxygen ions which in turn generates the Nernst voltage E . . . 18
2.9 A cross section showing the principal structure of the broadband lambda sensors 19 2.10 Cross section showing the principle for dual pump NOxprobes . . . 19
2.11 Mounting recommendations of the probe in an exhaust stream . . . 20
2.12 A simplified overview of the UV-RAS system . . . 21
2.13 Principle for dual beam NDIR CO2 detectors . . . 23
3.1 System overview of the final concept . . . 28
3.2 Cross section of design suggestion for multiple NOx probes . . . 29
4.1 Functional overview of the PEMSLight system. The external interfaces are shown as they are connected to the Main Unit . . . 31
4.2 Overview of the tailpipe attachment. The two modules are separated by a joint right after the pitot tube. Flow direction is from left to right in the figure . . . 33
4.3 Transparent overview of the tailpipe attachment. The two modules are sepa- rated by a joint right after the pitot tube. Flow direction is from left to right in the figure . . . 33
4.4 Overview of the NOx test set-up . . . 35
4.5 Flow rig set-up to compare the PEMSLight design of flow measurement against the Horiba design . . . 38
4.6 Flow rig set-up for the two cases of pipe endings in order to evaluate this influence on the Swema flow meter . . . 39
5.1 Low range step-up NOx test, 0-100 ppm NOx concentration (uncalibrated) . . . 41
5.2 High range step-up NOx test, 0-1000 ppm NOx concentration (uncalibrated) . . 42
5.3 Low range step-down NOx test, 100-0 ppm NOx concentration (uncalibrated) . 42 5.4 Low range step-up NOx test, 0-100 ppm NOx concentration (fusion) . . . 43
5.5 High range step-up NOx test, 0-1000 ppm NOx concentration (fusion) . . . 43
5.6 Low range step-down NOx test, 100-0 ppm NOx concentration (fusion) . . . 44
5.7 NOx sensor spread before and after linear calibration, NOx concentration dis- played for time reference . . . 45
5.8 Low range step-up NOx test, 0-100 ppm NOx concentration (calibrated) . . . . 46
5.9 High range step-up NOx test, 0-1000 ppm NOx concentration (calibrated) . . . 47
5.10 Low range step-down NOx test, 100-0 ppm NOx concentration (calibrated) . . . 47
5.11 Low range step-up NOx test, 0-100 ppm NOx concentration (calibrated fusion) 48 5.12 High range step-up NOx test, 0-1000 ppm NOx concentration (calibrated fusion) 48 5.13 Low range step-down NOxtest, 100-0 ppm NOx concentration (calibrated fusion) 49 5.14 Low range step in NOx (uncalibrated) . . . 49
5.15 Low range step in NOx (calibrated) . . . 50
5.16 Velocity test in controlled flow (unfiltered) . . . 51
5.17 Velocity test in controlled flow (low pass filtered) . . . 51
5.18 Cross correlation between PEMSLight CO2 and Horiba CO2 . . . 53
5.19 Cross correlation between PEMSLight flow and Horiba flow . . . 53
5.20 Cross correlation between PEMSLight NOx and Horiba NOx. . . 54
5.21 Comparison of the PEMSLight flow measurement with and without digital low pass filtering . . . 55
5.22 Comparison of uncalibrated PEMSLight volumetric flow measurement and mea- surement by Horiba . . . 55
5.23 Comparison of PEMSLight volumetric flow measurement and measurement by Horiba . . . 56
5.24 Linearity between Horiba and PEMSLight flow measurement for the entire test 56 5.25 Comparison of PEMSLight lambda based volumetric flow estimation and flow measurement by Horiba . . . 57
5.26 Linearity between Horiba and PEMSLight flow estimation based on lambda and fuel flow for the entire test . . . 58
5.27 Comparison of PEMSLight novel volumetric flow estimation and flow measure- ment by Horiba . . . 59
5.28 Linearity between Horiba and PEMSLight flow estimation based on O2 concen- tration and fuel flow for the entire test . . . 59
5.29 The uncalibrated flows from PEMSLight, measured with pitot and estimated with lambda, compared to Horiba . . . 60
5.30 Comparison of CO2 concentration calculated from O2 concentration and CO2 measurement by Horiba . . . 61
5.31 Linearity between Horiba and PEMSLight CO2 measurement for the entire test 61 5.32 Comparison of PEMSLight NOx measurement and measurement by Horiba . . 62
5.33 Close view of the 4 NOx sensor signals and Horiba signal . . . 63
5.34 Linearity between Horiba and PEMSLight NOxmeasurement for the entire test 63 5.35 PEMSLight NOx sensor O2 dependency. Values are scaled for comparability . . 64
5.36 Comparison between calculated NOx mass flow from PEMSLight and Horiba . 65 5.37 Linearity between Horiba and PEMSLight NOxmass flow measurement for the entire test . . . 65
5.38 Accumulated NOx comparison between PEMSLight and Horiba . . . 66
List of Tables
3.1 Evaluation of different flow measurement solutions . . . 25
3.2 Evaluation of different NOxmeasurement solutions . . . 26
3.3 Evaluation of different CO2 measurement solutions . . . 27
3.4 A table of the sensor in the final design concept . . . 28
4.1 Comparison of tested differential pressure sensors . . . 34
5.1 Calibration constants for the NOxsensors 2, 1, 4 and 3. . . 46
5.2 Results from investigating the influence on measured air speed due to the posi- tion of the Swema flow meter . . . 52
5.3 Table over the flow calibration constants used in the results . . . 60 5.4 Comparison of the accumulated NOx from the whole system test. The results
are presented as normalised values to the Horiba OBD-2200 measured NOxvalue 66
λ A measure of the air to fuel ratio in the exhaust ADC Analog to digital converter
CAN Controller area network CLD Chemiluminescence detector
CO Carbon monoxide
CO2 Carbon dioxide
Euro VI EU exhaust emissions legislation FID Flame ionization detector GND Electrical ground
GPIO Genereal purpose input/output GPS Global position system
HTS High temperature sensor NDIR Nondispersive infra-red NOx Nitrogen oxides
OBD On board diagnostics
PEMS Portable emissions measurement system PM Particulates or particle mass
ppm Parts per million PSU Power supply unit rpm Rounds per minute
SCR Selective catalytic reduction THC Total hydrocarbons
UV-RAS Ultra-violet resonant absorption spectroscopy VCC Positive power supply pin
ZrO2 Zirconia
1 Introduction
1.1 Background
From the 1st of January 2014 emissions of newly produced heavy duty vehicles have to meet the formal requirements set by the European parliament, known in the industry as Euro VI.
The equipment used for the certification process is regulated and has to meet certain criteria.
These criteria include which emission gases to measure as well as the detector type to use for the different gases. The strict nature of these requirements greatly limit the choice of measurement equipment.
Aside from the tests performed for the certification process, emission testing is performed continuously during the development of new products at Scania CV AB. Currently the system used for emissions testing at Scania is the Horiba OBS-2200. It uses a customised tailpipe to sample and then lead the exhaust gases via a heated tube to the main unit which is located in the cabin of the vehicle being tested.
In addition to its own sensors the system samples the OBD, on board diagnostics, of the vehicle being tested for torque, rpm, fuel consumption, speed, and temperatures. During tests the system outputs the values of CO, CO2, NOx and THC, total hydrocarbon, in real time at 1-10 Hz.
1.2 Problem Description
There are several problems with installation and operation of the currently used system.
As parts have to be removed and equipment need to be installed in the cab as well as under the vehicle, the set-up time is over two hours for an experienced mechanic. However if the equipment first has to be removed from another vehicle the time spent on installation is three to four hours.
Once the equipment has been installed it needs to be warmed up to operating temperature before any measurements can be made. The heated tubing needs to reach 191 ℃ to prevent water vapour in the exhaust from condensing. Additionally the hydrocarbon detector needs to reach 191 ℃ and the nitrogen oxide detector needs to be heated to 60 ℃. The heating process can require up to two hours, depending on ambient temperature, before reaching the appropriate temperatures.
Beside long set-up and start-up times the Horiba OBS-2200 consumes 0.5 kilowatts continu- ously thus requiring an external diesel power plant. The system also requires two additional calibration gases beside the combustion gas for the hydrocarbon detector.
1.3 Purpose & Goals
The purpose of this master thesis project is to develop a new emission analysis system to sim- plify the measurement process. The main focus is the ability to evaluate emission performance of a test vehicle during development, not necessarily meeting the official EU-regulations.
The purpose has been divided into the following goals:
• To research problematic areas and possible improvements on current emission measure- ment system at Scania CV AB to produce criteria for evaluation of a new system.
• To research different solutions solving these problems and critically assess them in rela- tion to how they correspond to the criteria.
• To conceptualise a mechatronic emissions analyser based on found solutions.
• To verify the concept and provide a comparison with the current emission measurement system as well as evaluate it in relation to the established criteria.
1.4 Limitations
This project is limited to using commercialised, or semi-commercialised, sensors, that is to say no new sensor types will be developed. The project does not aim to follow EU-regulations regarding heavy duty vehicle certification and the approved detector types. The project is intended to provide a customised emission measurement system for Scania CV and will be developed with regard to their specific needs. The report can not be considered an exhaustive search. The prototype is aimed to be used on a moving heavy duty vehicle, not in test cells. It is not included in the project to develop post-processing software to calculate emissions using
"work based windows", only to provide the same data as the Horiba system.
1.5 Methodology
1.5.1 Project Structure
The following steps of the project were identified:
1. Pre-study
2. Literature review 3. Prototyping 4. Verification
It was decided that the first stage of the project was to perform interviews with the involved engineers as a pre-study. This was to allow for both a qualitative, focusing on the users of the system, as well as a quantitative evaluation of the final prototype. The interview was designed in such a manner that important problems and improvements to the current system can be identified. To properly formulate the questions a short review of the current system had to be carried out. The result from the interviews was also intended to be used to weigh and prioritise the literature review so that important areas were highlighted and evaluated.
The literature review was expected to give a deeper knowledge of the subject and discover problematic areas before the actual development of the prototype. It was also to gather infor- mation on current emission measuring techniques available, resulting in a chapter containing the current state of the art, and make way for educated decisions as to which solutions to be used in the development of the prototype.
The development of the prototype was to provide a tool to evaluate the performance of the chosen solutions in comparison to Scania’s current measurement system and to see whether it would fulfil the criteria set by the interviews.
In the verification part of the project it was included to both test against the currently used system, Horiba OBS-2200, and to test the modules separately to verify their individual func- tion.
1.5.2 Rejected Alternative Methods
At an early stage in the project the possibility of starting with the literature review as early as possible was discussed, thus bypassing the interviews. This was however discarded due to the risk of having a too wide and unfocused literature review. The interviews give a deeper insight into which parts of the prototype is crucial to Scania and an indicator of what can be left out.
Concerning the concept evaluation it could possibly have been carried out using models and simulations instead of building a prototype, but then it would have been harder to provide a comparison with the existing system. Also the performance would have been harder to evaluate as some simplifications are always made when constructing models, resulting in the models not reflecting a real life prototype.
2 Frame of Reference
2.1 Interview
2.1.1 Design
The questions where formulated in conjunction with a Scania supervisor and based on a short background study. The interview questions were divided into sections regarding the problems of the currently used equipment, possible missing features and improvements, current usage, what is measured and how the measurements are used.
2.1.2 Compilation of Answers
Interpretation
The questions regarding problem with the current equipment revealed that most users had problems with the equipment being too large, including both gas bottles and a diesel generator.
This leads to both lengthy installations, requiring passenger seat removal, as well as problems finding space for the measurement equipment in the passenger compartment. Also the need of three different gas bottles in the cab was a problem for many users.
Aside from problems related to installation many users felt that the time spent waiting before tests, for the equipment to heat up or to calibrate itself, was too long and very tiresome. It was said to take up to 4 hours for each installation, with an additional heat up time of around 1 hour before the test can be started.
A key area of concern was also identified, namely that the current equipment can have large drifts in the zero-level for measurements of NOxand CO2. For NOxthe zero level can drift over 1% during a test, thus being a rather big error source when measuring in low concentration ranges.
The interviews also showed that a sensor accuracy of measuring NOx should be prioritised over the other Euro VI regulated emissions.
According to most of the users the order of priority was:
1. Ease of installation 2. Ease of use
3. Low zero-drift
As of additional improvements there were suggestions of also being able to measure NH3
(ammonia), doing real-time calculations of mean-values, displaying momentary measurement values during the test as well as displaying the so far accumulated NOx concentration. It was also suggested that the prototype could be designed to be connected to one of the internal CAN-buses, making it possible to include even more of the vehicle’s sensory data instead of just the ones available on the OBD-bus.
Regarding measurement performance most of the interviewed users mentioned a required span of 0-1500 ppm and extra accuracy below 50 ppm for the NOx sensor.
From the answers collected during the interviews it could be gathered that the prioritisation of the sensor capabilities of the finished prototype would be as follows:
1. NOx sensor 2. Gas flow sensor 3. CO2 sensor 4. NH3 sensor 5. GPS module 6. PM sensor
This list was to be used when prioritising which sensor to implement in the concept and later in the prototype.
Evaluation Criteria 1. Installation time 2. Ease of installation
• Need of external power
• Need of gasbottles
• Size
• Need of sampling tube from exhaust 3. Startup time
• Heat up time
• Calibration time 4. Accuracy
5. Low zero drift 6. Sampling speed
2.2 Horiba OBS-2200
One of the current measurement systems at Scania is the OBS-2200 by Horiba. It is a On- Board System, capable of measuring concentrations of CO, CO2, THC (total hydrocarbon) and NOxwhile being mounted in the vehicle itself. This enables on-road measurements of the vehicle while its driving. The system can also be equipped with GPS for global positioning, a flow meter to measure the exhaust flow and a weather station to gather ambient data during the tests. The system provides the ability to collect data from the OBD-bus if available and calculate mass emissions if the flow meter is installed.
The Horiba system uses several different sensors to detect the different emissions. The CO and CO2 is measured using a NDIR (nondispersive infra-red sensor), THC by a FID (flame ionization detector) and the NOxwith CLD (chemiluminescence detector). A heated sampling
tube is used to extract exhaust gases from the mount on the tail pipe into the main system-box as illustrated by Figure 2.1.
Tailpipe attachment
Main unit Heated sample
line Typical system setup on a truck with trailer
seen from above
Figure 2.1: Example of mounting the Horiba system in a truck
The system consists of the tail pipe attachment, the main system box and the included laptop as can be seen in Figure 2.2. The main system box is approximately 500 mm deep, 350 mm wide and 330 mm tall with a weight of around 29 kg.
350mm
330mm 500mm
Main unit
Tailpipe attachment
Heated sample line
Gastubes
Horiba OBS-2200 System overview
Figure 2.2: Simplified overview of the Horiba system
2.2.1 Performance
Since the resulting prototype of this project was to be compared to the Horiba system used at Scania the performance of the OBS-2200 is of great interest. The source of the information in this section is Horiba OBS-2200 Manual (2010).
NOx
The Horiba system is capable of measuring, using CLD, NOx concentrations in several ranges when in a mode called "auto range", enabling switching between measurement ranges during run time, and can do this with a accuracy of ±2.5% of the current maximum limit in the range. The ranges available are 0-100 ppm, 0-500 ppm, 0-1000 ppm and 0-3000 ppm.
This means that, if the measured NOx concentration is within the lowest available range, an accuracy of ±2.5 ppm can be achieved. As for how the accuracy of the system was deter- mined Horiba refers to CFR 1065.602 in their instruction manual for the OBS-2200, where the
accuracy is defined as
accuracy = |1 N
N
X
i=1
(yi− yref
i)|. (2.1)
Additionally, as stated in Compilation of Answers at page 5, a rather significant zero-level drift has been observed by the users. The drift, as stated in the instruction manual, is to be within ±2.0% per 4 hours at an ambient temperature fluctuating within ±2 ℃. There is also a temperature-dependent drift regarding ambient temperature, determined to ±3% every 10
℃ during ambient temperatures of 0-40 ℃.
CO2
The system can measure, using NDIR, the CO2 volume concentration in the exhaust gases within a range of 0 − 20%, with a accuracy of ±2.5% of the current max range.
Flow
The flow is being measured with a two point averaging pitot tube, achieving an accuracy of
±1.5%of maximum range or ±2.5% of the current measurement reading, using the definition of accuracy as in Equation 2.1. The range is dependant on the tail pipe diameter and the exhaust flow rate, with a maximum range of 0-65 m3/min.
2.3 Gas Flow Measurement
2.3.1 Problem Description
The one dimensional flow through a pipe section, see Figure 2.3, can be related to the mass flow ˙m and volume flow Q according to Equation 2.2 and Equation 2.3.
˙
m = ρAu (2.2)
Q = Au (2.3)
where u is the flow velocity, A is the pipe inner area and ρ the density of the gas.
Flow D
L
ρ, p, T
Figure 2.3: Cross-section of a pipe showing flow direction and dimensions
The density ρ of the gas relates to the pressure of the gas p and the temperature of the gas T according to the ideal gas law, see Equation 2.4. It can be modified to express the change in density caused by variations in pressure and temperature, see Equation 2.5.
p = ρRT (2.4)
p1 p2
×T2 T1
× ρ2 = ρ1 (2.5)
where R is the specific gas constant of the gas.
Flow measurement is necessary in order to calculate the mass flow of different substances from the measured concentrations in the exhaust.
2.3.2 Solutions Pitot Tube Flow Meter
The pitot tube based flow meter functions on the principle of measuring both the stagnation pressure p0 and the static pressure p of a fluid or gas and using Bernoulli’s equation to calculate the speed, see Equation 2.6.
p0= p + 1
2ρu2= const (2.6)
The stagnation pressure is often measured using a pitot tube, hence the name of the meter type, see Figure 2.4a. Other types of measurement tubes exist such as the Prandtl tube and the averaging pitot tube, see Figure 2.4c and Figure 2.4b. One important aspect of the pitot tube method is that it provides a measurement of the flow velocity in one point, not the mean velocity in a pipe section.
Flow D
Stagnation pressure tube Static pressure tube
ρ, p, T
(a) Pitot tube in pipe
Flow D
Static pressure tube Stagnation pressure tube
ρ, p, T
(b) Average pitot tube in pipe
Flow D
Static pressure tube Stagnation pressure tube
ρ, p, T
(c) Prantl tube in pipe
Figure 2.4: Figures of different pitot tubes
By expanding Equation 2.6 and solving for the flow velocity u, Equation 2.7 is produced.
As clearly seen in the equation the flow velocity is proportional to the root of the pressure difference.
u = s
2∆p
ρ = c1p
∆p (2.7)
where ∆p = p0− pand c1 =q
2 ρ.
The non-linear relation of u to ∆p can cause some problems. Consider the arithmetic mean of u, for 2 values of u it would be defined as Equation 2.8.
u = u1+ u2
2 (2.8)
Inserting Equation 2.7 into Equation 2.8 produces Equation 2.9, assuming the density is the same.
u = c1√
∆p1+ c1√
∆p2
2 = c1
2
p∆p1+p
∆p2
= c2p
∆p1+p
∆p2
(2.9)
where c2 = c21 =
q2 ρ
2 =q
1 2ρ.
Now consider the arithmetic mean pressure difference, ∆p, calculated according to Equa- tion 2.10.
∆p = ∆p1+ ∆p2
2 (2.10)
Inserting Equation 2.10 into Equation 2.7 produces Equation 2.11, the calculated flow velocity v from an averaged pressure difference.
v = c1
q
∆p = c1
r∆p1+ ∆p2
2 = c1
√2
p∆p1+ ∆p2= c3
p∆p1+ ∆p2 (2.11)
where c3= √c1
2 =q
1 ρ.
By assuming u 6= v and inserting Equation 2.9 and Equation 2.11 one gets Equation 2.12.
c2
p∆p1+p
∆p2
6= c3p
∆p1+ ∆p2 (2.12)
If the right and left side of Equation 2.12 is squared the result is Equation 2.13
c22
∆p1+ ∆p2+ 2p
∆p1∆p2
6= c23(∆p1+ ∆p2) (2.13)
Dividing both sides of Equation 2.13 with c22 one gets Equation 2.14.
∆p1+ ∆p2+ 2p
∆p1∆p2
6= 2 (∆p1+ ∆p2) (2.14) Subtracting both sides with the left hand side of Equation 2.14 one gets Equation 2.15.
0 6= 2 (∆p1+ ∆p2) −
∆p1+ ∆p2+ 2p
∆p1∆p2
= ∆p1+ ∆p2− 2p
∆p1∆p2 (2.15) Factorising the right hand side of Equation 2.15 gives Equation 2.16. As the right hand side of Equation 2.16 is squared it can never become negative hence it is always larger than the left hand side given that the difference between pressure samples ∆p1 and ∆p2 is not zero.
0 <p
∆p1−p
∆p2
2
(2.16)
The meaning of this is that if averaging between samples is done in the pressure domain, instead of the velocity domain, the calculated flow velocity will always be overestimated (Laurantzon, 2010; Nakamura et al., 2005). This holds for both samples averaged in time and in different locations. This means that measuring the mean stagnation pressure in a pipe section, as an average pitot tube does, and calculating the corresponding flow velocity is not the same as measuring the average flow velocity in the pipe section.
Ultrasonic Flow Meter
When discussing the ultrasonic flow meter it is important to distinguish between different types, all of them ultrasonic but based on different principles. The methods worth to mention are the "Transit-time flow meter", the "Doppler meter" and the "Cross-correlation flow meter"
(Baker, 2000, ch. 13). In this part only the transit time flow meter will be explained and discussed as the applicability of the other two, for different reasons, is questioned. Firstly the Doppler meter requires "objects" to travel with the flow to provide a surface for which ultrasonic waves can bounce off of(Baker, 2000, ch. 13), something not present in the flow of exhaust gas. Secondly the cross-correlation meter requires heavy turbulence, bubbles or other large fluctuations in the gas flow to function properly, it also requires electronics able to compute the cross-correlation of the ultrasonic receivers in real time making it an expensive method (Baker, 2000, ch. 13).
The transit-time method uses the difference in transit-time between sound waves propagating with and against the flow of the propagation medium. Sound waves travelling downstream will have a shorter transit-time and sound waves travelling upstream will have a longer transit-time, given a positive flow velocity, see Equation 2.17 and Equation 2.18.
tdownstream= L
c + u cos β (2.17)
tupstream= L
c − u cos β (2.18)
where c is the speed of sound in the medium, u is the flow velocity, L is the distance between the ultrasonic transducers and β is the angle between the transducers.
The ultrasonic transducers need to positioned along the length of pipe, often at opposite sides of the pipe, and the sound has to be sent at an angle β 6= 90 deg so that there is a component of the fluid velocity along the path of the acoustic beam, see Figure 2.5.
Flow β D
L Ultrasonic transducers
ρ, p, T
Figure 2.5: Geometrical overview of the transit-time flow meter
The flow velocity in the pipe section can be obtained by combining Equation 2.17 and Equa- tion 2.18, see Equation 2.19.
u = L
2 cos β
1
tdownstream
− 1
tupstream
= D
2 cos2β
1
tdownstream
− 1
tupstream
(2.19)
As seen in Equation 2.19 the flow velocity is only dependant on the transit-times, the angle between sensors and the acoustic beam path, hence the speed of sound in the medium is not needed. However because the distance of travel is so short and the speed of sound in gas so high the transit-times would be in the microsecond range, thus requiring fast electronic time measurement processors to gain reasonable precision in velocity measurements (Baker, 2000, ch. 13).
The transit-time flow meter automatically averages the flow velocity along the acoustic beam path, thus compensating for some variations in flow. However as it only averages along the beam path the measured velocity is not always the same as the average in the cross section of the flow. If the flow is fully laminar the flow velocity will be zero at the pipe edges and reach its maximum in the middle of the pipe, see Figure 2.6. The measured flow can be 33% higher than the mean flow due to the averaging along the beam path (Baker, 2000).
Flow D
L
ρ, p, T
(a) Side view of pipe
Flow D
(b) Front view of pipe Figure 2.6: Laminar flow in pipe section
Flow Estimation Using λ and Fuel Consumption
One method designed to get around measuring the exhaust flow is the estimation using intake flow according to Equation 2.20. The basic idea is that all mass flowing into the engine must flow out after a time delay.
Xm˙out(t) =X
˙
min(t − tdelay) (2.20)
The mass flows in to the engine can be divided into ˙mair and ˙mf uel, while the only mass flow out of the engine should be ˙mexhaust leading to Equation 2.21.
˙
mexhaust(t) = ˙mair(t − tdelay) + ˙mf uel(t − tdelay) (2.21) The volumetric flow of fuel into the engine is already available as part of the on board diag- nostics (OBD), from which the mass flow of fuel can be calculated. The mass flow of air can be calculated from ˙mf uel and from the air to fuel ratio, see Equation 2.22 (Heywood, 1988, chap. 3).
AF R = mair
mf uel (2.22)
The theoretically ideal ratio of air to fuel, with enough oxygen to react with all of the fuel, is called the stoichiometric air to fuel ratio. If the air to fuel ratio is less than the stoichiometric ratio there is not enough oxygen to react with all the fuel, called rich burn, and if over the stoichiometric ratio, called lean burn, the extra oxygen appears unchanged in the exhaust (Heywood, 1988, chap. 3). λ is a measure of how close to the stoichiometric ratio the air to fuel ratio is, it is defined as in Equation 2.23.
λ = AF R
AF Rstoich (2.23)
AF Rstoich is calculated using the complete combustion equation for hydrocarbon fuel, Equa- tion 2.24. As diesel is a mixture of hydrocarbons, ranging from C10H20 to C15H28, and other compounds the equation cannot fully reflect reality.
CaHb+
a + b
4
(O2+ 3.773N2) → aCO2+ b
2H2O + 3.773
a + b
4
N2 (2.24) By inserting the longest hydrocarbon for diesel one gets
C15H28+ 22(O2+ 3.773N2) → 15CO2+ 14H2O + 83N2. (2.25) To obtain the AF Rstoich the molecular mass of air, mmolecularair = 138.25, and C15H28, mmolecularC15H28 = 208.39, as well as the masses of the products must be inserted for the mentioned items, thus producing Equation 2.26.
208.39(f uel) + 22 × 138.25(air) → 15 × 44.01(CO2) + 14 × 18.02(H2O) + 83 × 28.16(N2) (2.26) Extracting the total mass of air and fuel and inserting them into Equation 2.22 the value of AF Rstoichcan be calculated, see Equation 2.27 (Heywood, 1988, chap. 3).
AF Rstoich= 22 × 138.25
208.39[g] = 3041.5[g]
208.39 = 14.60 (2.27)
By using a lambda sensor in the exhaust, λ can be directly measured and therefore the air to fuel ratio can be calculated using Equation 2.27 and Equation 2.23, which was needed to calculate ˙mairfrom ˙mf uel. Combining all equations the equation for ˙mexhaustcan be extracted, seen in Equation 2.28.
˙
mexhaust= (AF Rstoichλ + 1)Qf uel
ρf uel (2.28)
An exhaust measurement system based on the principle of flow estimation using intake air and fuel was tested and compared to a more traditional CVS-based approach by Kihara et al. (2000). The measurement system produced NOx results within 4% of the reference system.
The time delay between the inflow and the outflow of mass can be predicted and modelled using engine manifold and pipe filling and emptying (Steigerwald et al., 2007). It can also be ignored with unknown consequences to emission calculations.
A study on error propagation in gas flow measurements by Olsson et al. (2013) discusses the possibility to use lambda measured in the exhaust to calculate gas flows through the engine, including ˙mexhaust, but in the end advises against it for lambda above 1.5. It is instead recommended to use sensors to accurately measure the gas flows.
2.4 NO
xMeasurement
2.4.1 Problem Description
As deduced from the interviews the most important feature to be included in the prototype was the ability to measure the NOx(NO+NO2) of the vehicle exhaust. There are several ways of measuring NOx and the currently used standard in Europe is the Chemi-Luminescence detector. This is also the NOx detector type of the Horiba OBS-2200 measurement system currently in use at Scania CV AB.
The NO usually forms from atmospheric nitrogen together with oxygen at high temperatures and it has been found that most of the NO forms in the engine after the actual flame, due to the additional rise in pressure and temperature. According to Heywood (1988) the formation of NO from atmospheric nitrogen can be explained by Equation 2.29, Equation 2.30 and Equation 2.31.
O + N2 → N O + N (2.29)
N + O2 → N O + O (2.30)
N + OH → N O + H. (2.31)
2.4.2 Solutions
Chemi-Luminescence Detector (CLD)
Chemi-luminescence is the phenomenon of light being radiated from a chemical reaction. The principle is that two, or more, substances react with each other, resulting in a product in which a fraction of it, about 10%, is in a electronically excited state. This new excited substance then progresses to the ground excitation state, emitting the excess energy as photons (Baeyens et al., 1998; García-Armingol et al., 2013; Baronick et al., 2001) according to Equation 2.32.
A + B → C∗ → C + light (2.32)
where C∗ is the excited radical of C.
This phenomenon is the foundation of the CLD as the light emitted can be detected and analysed in an attempt to determine the originating substances.
In the specific area of sensing NOx the reaction between NO and O3 (ozone) is utilized, resulting in electronically excited NO2 and oxygen molecules (Simmons and Seakins, 2012;
Navas et al., 1997).
N O + O3→ N O∗2+ O2→ N O2+ O2+ light (2.33)
Reaction chamber Exhaust
gases
Photo detector
Ozone
Vacuum pump
Catalyst
NO O3 NO + NO2
Figure 2.7: Overview of the principle behind a chemiluminescence system
To determine the concentration of NO in the exhaust stream a sample can be extracted trough a tube into a reaction chamber and combined with a excess amount of ozone. The emitted light, under the correct circumstances, is then proportional to the amount of NO present in the sampled gas. This does however not cover the NO2 present in the sample, but solutions to this problem exists in the form of using e.g. a catalyst to convert the NO2 to NO before the reaction with ozone as can be seen in Figure 2.7 (Simmons and Seakins, 2012).
Generally the advantages of this method is that it is relatively sensitive and can measure low concentrations as well as using simple and inexpensive equipment (Navas et al., 1997).
As for disadvantages there have been some problematic interference reported for measuring NOx in combustion engines due to other substances present in the sample. Molecules such as H2O, CO2 and O2 has been proven to influence the measured NOx. In the ideal case Equation 2.32 and Equation 2.33 are valid, but in other molecules presence the energy emitted could also be absorbed by said molecule as seen in Equation 2.34 (Matthews et al., 1977).
C∗+ M → C + M (2.34)
Zirconia Based Electrochemical Sensor
For in-situ measurement a lot of the available high performance measurement methods, such as CLD, falls short in terms of user friendliness and is usually complex, sometimes requiring additional external resources to function. The measurement on actual vehicles on road raises problems such as chemical contaminants, high temperature and other harmful environments for sensitive equipment (Szabo et al., 2003).
A sensor type that shows promising NOxsensing capabilities in these harmful environments is the solid-state sensor made of zirconia (ZrO2). There are several different sensor types based on this type of sensor, as explained by Fleming (2001).
These zirconia, or zirconium oxide, based sensors work by a principle known as early as 1889, described by the so called Nernst equation, shown in Equation 2.35 for oxygen. The ZrO2
is formed into an electrochemical cell, aka Nernst cell, which reacts to the partial pressure difference in oxygen between the two electrodes at the two sides of the cell which in turn generates a Nernst voltage, as can be seen in Figure 2.8 (Riegel et al., 2002).
Zr
Electrode
Electrode
O2- O2
O2
O2
O2 O2
O2 O2 O2-
O2- O2-
O2 E
p1
p2
e- e-
e-
Figure 2.8: Nernst cell principle showing the transfer of oxygen ions which in turn generates the Nernst voltage E
The Nernst equation for oxygen is
E = − RT 4F
p2O2
Z
p1O2
tion
pO2
dp (2.35)
where E is the galvanic potential, pxO2 is the partial pressure of oxygen at side x, tion is the ionic transference number for oxygen ions, R is the gas constant and T is the absolute
temperature.
What further simplifies this equation is tion' 1 for zirconia at higher temperatures, meaning that almost all of the conductivity is by ionic conductivity and none by electrical (Eddy, 1974).
This leads to the following equation
E = − RT 4F
ln p1O2 p2O2
. (2.36)
One must note that for Equation 2.36 the zirconia must reach a temperature of around 400
℃ or higher in order to reach the assumed ionic conductivity of tion ' 1.
A number of variations of these zirconia (ZrO2) based sensors are mentioned by Fleming (2001). The earliest versions mentioned being capable of producing a step-like output at the stoichiometric point and thus only being able to tell rich and lean burns apart. The newer models, like the broadband lambda probe, can with increased precision and range even measure what lambda value the engine runs at with a precision down to a couple of ppm of oxygen concentration.
These sensors adapts a dual chamber technique, having structure with a reference chamber and a measurement chamber.
Exhaust flow
Oxygen reference chamber
Pumping cell electrodes Diffusion passage
Rerefence electrodes Measurement
chamber
Figure 2.9: A cross section showing the principal structure of the broadband lambda sensors
As can be seen in Figure 2.9 one of the chambers is open to the gas that’s being measured (i.e.
the measurement cell), allowing a gas exchange in and out of the chamber, and the other one is a reference chamber, containing a known oxygen concentration. There are two pairs of electrodes in this structure, one for pumping oxygen in/out of the measurement chamber and one for comparing the oxygen concentration in the measurement chamber to the reference.
The pump cell is used to keep the air-to-fuel ratio of the measurement chamber at the so called stoichiometric ratio (λ = 1) while under the continuous external disturbance of the gas exchange. The current the pump cell consumes in the process of keeping λ = 1 is then proportional to the oxygen concentration of the exhaust. The reference cell is used to compare the measurement chambers oxygen concentration to a known value, thus allowing for a accurate regulation of the measurement chamber.
The zirconia NOxsensor is similar to the dual chamber design, shown in Figure 2.9. The main difference is that the NOx sensor uses an additional pumping cell together with an additional chamber (Riegel et al., 2002).
Exhaust sample
Oxygen
Oxygen NOx Oxygen
reference cell
Electrodes
Diffusion passage
Figure 2.10: Cross section showing the principle for dual pump NOx probes
As seen in Figure 2.10 the first chamber reduces the oxygen level in the gas. This is done in the same as in the broadband oxygen sensor, i.e. by keeping the gas in the chamber at
the stoichiometric point, λ = 1, by the pumping of oxygen. This function also enables some of these sensors to give the current lambda value as well as the NOx value, acting as both a lambda probe and a NOx probe. The gas then moves through an diffusion gap down to the NOx sensing chamber which pumps the remaining oxygen out of the chamber, reducing the NOx to N2 and O2. The current used in this extra pumping cell is then proportional to the amount of NOx in the chamber (Inagaki et al., 1998).
Cross sensitivity towards NH3 has been observed. The NH3 oxidises in the sensor’s first chamber and can then be mistaken for being initial NOx emissions (Frobert et al., 2013), as can be seen in Equation 2.37, Equation 2.38 and Equation 2.39.
2N H3+ 2O2 → N2O + 3H2O (2.37)
4N H3+ 5O2→ 4N O + 6H2O (2.38)
4N H3+ 7O2 → 4N O2+ 6H2O (2.39)
This is a known problem with the zirconia sensors, which makes the use of them in systems with SCR (selective catalytic reduction) difficult if the system suffers from much ammonia slip.
Another thing that needs to be considered when operating the probe is the high temperature required for it to function. This causes the sensor to be extra sensible to water condensation as this can completely break the sensor when heated. Therefore it is usually recommended to mount the probe in an upright position, in order to prevent water from collecting in the sensor, as shown in Figure 2.11.
Exhaust flow
< 80 deg
Exhaust pipe
< 10 deg
< 10 deg
Figure 2.11: Mounting recommendations of the probe in an exhaust stream
UV Spectroscopy Analyser (UV-RAS)
The UV-RAS (ultra-violet resonant absorption spectroscopy) described by Heller et al. (2004) uses the absorption of specific wavelengths within the UV-spectrum in exhausts to identify the concentrations of NO and NO2 in a built in sample cell, illustrated in Figure 2.12.
UV source
Filter Beam splitter
Sample cell
Detector
Reference detector Exhaust sample
Figure 2.12: A simplified overview of the UV-RAS system
The wavelengths of the UV-radiation are chosen with respect to the specific absorption prop- erties of NO and NO2, thus avoiding unnecessary interference from other compounds present, as is the case with exhaust gases. The source of the ultra violet radiation is a electrode-less discharge lamp (EDL). The lamp uses a mixture of oxygen and nitrogen which together with plasma, created by the electrode-less discharger, produces excited NO, see Equation 2.40, Equation 2.41 and Equation 2.42.
N2+ O2 → 2N O∗ (2.40)
N O∗ → N O + light (2.41)
2N O → N2+ O2 (2.42)
The radiation from the reaction in Equation 2.41 is at a wavelength of 226.5 nm. There is also radiation emitted from excited nitrogen in the range of 400-700 nm, as can be seen in Equation 2.43.
N2∗→ N2+ light (2.43)
The emitted UV-radiation is further filtered before separated in a beam splitter, one part going to a detector and used as a reference and the other going through a sample cell with exhaust gas and into a detector.
This method experiences the similar drawback as the CLD of other contaminants affecting the readings. In the CLD this was mostly because of the quenching effect from other molecules, absorbing the energy from excited NO-molecules. In the UV-RAS the interference from other molecules comes from them absorbing the UV-radiation. One must though note that the reported effect on the UV-RAS is significantly lower (Heller et al., 2004).
The response time of the UV-RAS is said to be similar to the one of the CLD. A heated sampling tube is required if the gas sample has to be transferred from the measured object to avoid reaction of NO2 with other substances, such as H2O (Heller et al., 2004).
2.5 CO
2Measurement
2.5.1 Problem Description
When evaluating exhaust emissions of heavy duty vehicles a measurement based on the concept of "pollutant per kWh", as for NOxin [g/kWh], is most often used. This provides a good tool for evaluating emission levels based on engine power output. However when evaluating exhaust emissions there is another method based instead on the energy consumption of the engine,
"pollutant per mass CO2". The main benefit of this method is that it provides a tool for evaluating emission performance at low or no engine output, as when idling (Vermeulen et al., 2012). As CO2 emission of heavy duty vehicles is not regulated in EU-legislation (European Commission, 2011), CO2 is currently measured for calculating the theoretical fuel consumption using carbon balance.
2.5.2 Solutions
Oxygen Based CO2 Estimation
In a diesel engine the emission of CO2 is due to combustion of diesel fuel in the presence of oxygen, normally in the form of O2 as part of the earth’s atmosphere. During combustion the oxygen in the air is combined with the fuel to produce CO2 and H2O in the exhaust, see Equation 2.24.
The study by Schrenk and Berger (1941) shows a linear relationship between CO2 in the exhaust of a diesel engine and fuel to air ratio. As fuel to air ratio is the inverse of AF R, see Equation 2.22, this relationship can be used to calculate CO2 in the exhaust by measuring λ or oxygen instead (Vermeulen et al., 2012). The linear relationship is valid for 16.64 ≤ AF R ≤ 100or 1.14 ≤ λ ≤ 6.85. For lower AF R or λ the relationship is less linear but can still be used as an approximation, down to AF Rstoich or λ = 1. The coefficients of the linear relationship depends on intake oxygen content and fuel hydrogen and carbon content (Vermeulen et al., 2012). The equation used by Vermeulen et al. (2012) for calculating the CO2 concentration CCO2 from the O2 concentration CO2 can be seen in Equation 2.44.
CCO2 = −0.6463CO2+ 13.54 (2.44)
Nondispersive Infra-red Detector
The nondispersive infra-red detector, abbreviated NDIR, is a gas measurement method based on Lambert-Beer law, Equation 2.45.
I = I0e−kcl (2.45)
where I is the intensity transmitted light, I0 the incident light, l the length of light travel, k the gas absorption coefficient and c the gas concentration.
An infra-red light source continuously emits light of intensity I0 into a sample chamber where a gas sample is present, see Figure 2.13. The gas decreases the light intensity according to Lambert-Beer law and the resulting intensity I is measured by a detector behind a narrow bandpass filter that is selecting the wavelength characteristic to CO2. The incident light intensity I0 is measured by a separate detector with a bandpass filter selecting a wavelength without absorption by gases typically present in the gas sample (Nakamura et al., 2002). This type of NDIR detector is called dual beam detector and the intensity ratio II0 is equal to the ratio of the measured voltages UU0 (Schilz, 2000), see Figure 2.13.
V V
CO2
CO2 CO2
CO2
Other gas
Other gas
Other gas Other
gas
CO2 wavelength filter
Ref wavelength filter Sample gas in Sample gas out
IR source
U
U0
Detectors
l
Figure 2.13: Principle for dual beam NDIR CO2 detectors
A problem with using NDIR as a measurement method to measure CO2 in diesel exhaust gas is the interference of H2O. This problem is discussed by Nakamura et al. (2002) and a solution is proposed and tested.
3 Concept
3.1 Comparison and Evaluation
3.1.1 Flow Measurement
The different concepts for measuring flow, presented in Section 2.3, where evaluated according to the criteria from Section 2.1, a compilation can be seen in Table 3.1.
Table 3.1: Evaluation of different flow measurement solutions
Concept: Pitot Ultrasonic λ estimation
Installation time: Medium Medium Medium
Ease of installation
Need external power plant: No No No
Approximate power consumption: Low Medium Medium
Need of gas bottles: No No No
Size: Small Large Small
Need of sampling tube: No No No
Startup time
Heat up time: Zero Zero Low
Calibration time: Zero Zero Zero
Accuracy: Good Good Unknown
Drift: Low Low Unknown
Sampling speed: 2 kHz 500 Hz OBD dependant
All of the investigated flow measurement techniques require the installation of sensors in the exhaust, thus the installation time is about equal for the sensors if each of them are fitted to exhaust pipes ready to be mounted. On the other hand the lambda based estimation requires fuel flow which is only available as part of the the on board diagnostics and as a result it requires the installation of a component for reading the OBD-bus.
The different techniques all require a supply of external power, but none can be considered to require an external power plant. The power demand of pitot flow meter is considered to be the lowest as it requires neither heating nor cooling and the only thing requiring power is the pressure sensors. The ultrasonic solutions on the market have in common that they require cooling to function properly when measuring exhaust flow as it is a high temperature environment of several hundred degrees Celsius. The lambda based estimation technique on the other hand requires the lambda sensor to be heated and thus having a certain start-up time before being able to output values.
Because it requires cooling the ultrasonic equipment is in general a bit larger than with the other two methods. As mentioned previously all of the methods are mounted directly on the exhaust pipe. Only the estimation method is delayed by a heating time, as mentioned previously, the other two can start measurements immediately.
As the estimation method is highly dependant on the accuracy of both the lambda probe and the fuel flow measurement of the vehicle at hand it is hard to predict the actual accuracy of
the method, the drift is equally hard to predict because of the same reasons. The sampling speed of it is dependant on the update frequency of the on board diagnostics, which in turn is vehicle dependant.
Both the pitot meter and the ultrasonic meter have shown to have good accuracy and should have low drift. The sampling frequencies is in the range of hundreds of Hz and even kHz for the pitot meter (Nakamura et al., 2005; Beck and Hinterhofer, 1998).
3.1.2 NOx Measurement
Table 3.2: Evaluation of different NOx measurement solutions
Concept: ZrO2 NOx CLD UV-RAS
Installation time: Medium High Unknown
Ease of installation
Need external power plant: No Yes Yes
Approximate power consumption: 30 W 400 W 400 W
Need of gas bottles: No Yes No
Size: Small Big Big
Need of sampling tube: No Yes Yes
Startup time
Heat up time: Low High Unknown
Calibration time: Low High Low
Accuracy: ±10 − 15ppm ±2.5ppm ±0.04ppm
Drift: Low Low Low
Sampling speed: High Medium Medium
To properly choose the best suited NOxsensor for the task the alternatives had to be evaluated and compared to each other in Table 3.2. The first and most important part about the prototype, with regard to the users, were identified to be the ease of installation, as stated in Section 2.1.2. The three alternatives, mentioned in Section 2.4, are very well suited for the task at hand when comparing the size requirements since they all could be fitted at the passenger side of the truck cabin.
The ZrO2 NOx probe seems to really differentiate from the other two sensors with respect to the need of sampling tubes, as neither the CLD or the UV-RAS are mounted directly on the exhaust tube. When dealing with NO2 emissions this can be especially inconvenient since the exhaust need to be kept above the dew point during the transportation to avoid reaction with condensed water. To keep such a high temperature over such a long sampling tube will require large amounts of power and most likely an external power-source. This results in an additional component to install when setting up the equipment in a truck. The ZrO2 NOx probe can be positioned directly at the exhaust pipe and thus not requiring a heated sampling tube.
One must note that this probe also needs heating to a high temperature when operating, but the part that need heating is only the sensor element itself, which is designed to be of small thermal mass and is partly heated by the exhaust. The heating power of the probe is in the size of tenths of what is required for a heated sampling tube and requires less than 30 watts of
power at normal operation, measured in bench top tests, thus not needing an external power source other than the truck’s power outlet itself.
As it comes to heat up time the usual power on time of the ZrO2 NOx probe is relative small, but it is not possible to operate it in a truck before start up due to the large amounts of water vapour present in the exhaust. The sensor need to be started after the truck has started, thus not being able to sample the NOx at start-ups. In comparison the CLD is also affected by water in the sample, but countermeasures such as H2O-sensors has been observed to enable compensation of such disturbances, as presented in Section 2.4.
The CLD suffers from the need of being calibrated before each use, which is not the case for the UV-RAS or the ZrO2 NOx probe. This means that the CLD need to be accompanied by a bottle of calibration gas during the tests and is thus adding an additional component in the installation process.
3.1.3 CO2 Measurement
Table 3.3: Evaluation of different CO2 measurement solutions
Concept: NDIR O2 based estimation
Installation time: High Very low
Ease of installation
Need external power plant: Yes No
Approximate power consumption: High Medium
Need of gas bottles: No No
Size: Large Small
Need of sampling tube: Yes No
Startup time
Heat up time: Medium Low
Calibration time: Zero Zero
Accuracy: High Unknown
Drift: Low Unknown
Sampling speed: High High
While the NDIR method is the current standard for measuring CO2 in exhaust the O2 based method is relatively new and untested. On the other hand the installation of another sensor which requires a separate sampling tube will not provide any increased usability nor better performance as it is implemented in the currently used equipment. By contrast the O2 based method does not require a sampling tube as a lambda probe can be inserted directly in the exhaust pipe, it might not have the same performance but as it is a sensor that should not be prioritised the increased usability makes it an overall better choice.
3.2 Final Design
When combining the evaluations of all the different sensors and measurement methods a final concept emerged. A realisation was made concerning the combination capabilities of the
