Implementation options for ‘feed-back approach’ onroad-load determination
Author: Iddo Riemersma (expert for Transport & Environment)
Date:10 October 2012
Version: 1.2 (revised draft)
Background
This document is written in the context of the WLTP, the process that will develop a new world-harmonised test procedure and test cycle for the measurement of emissions of light duty vehicles. Goal of this WLTP process is to obtain a type approval test result that is more representative for real life, especially in terms of fuel efficiency/CO2 emissions. The current NEDC testprocedure used in the EU falls short in the measured fuel consumption by about 20%, due to flexibilities in the procedure and an unrepresentative test cycle [Mock 2012][Smokers 2012]. The determination of the road-load of a vehicle is an important contributor for this gap,and for that reason it receives much attention in the development of the new test procedure. Road-load is used as an input to set the load on the chassis dynamometer for the emissions test. It has a direct effect on the measured pollutants and the level of CO2 emissions/fuel consumption [Honkanen 2012]. The Swedish Transport Administration and the Belgian NGO TransportEnvironment have acknowledged earlierin the WLTP process the relevance of improving the road-load determination procedure, and provided a document as input for the discussion [Smeds 2011]. At the time there was little evidence available, but it was clear that the difference between the actual road-load and the one measured at type approval could lead to 15% higher CO2 emissions or even more. Recently TransportEnvironment and the Dutch Ministry for Infrastructure and Environment conducted a road-load verification project on 6 Euro 5/6 vehicles, which showed that the increase in CO2 due to a non-representative road-load is on average 11% [Kadijk 2012]. Recent tests on 6 vehicles in Finland for TM magazine showed a difference between reported and measured CO2 emissions of 28% on average, when using the actual measured road-load [Honkanen 2012]. This evidence shows there is a clear need for an improved road-load determination procedure. WLTP document [Smeds 2011] listed 2 different approaches to achieve that objective:
- Feed-forward approach
By setting more requirements and tightening the tolerances it is tried to ensure that the future production vehicle will match the measured road-load during type approval as close as possible. - Feed-back approach
With respect to the vehicle specifications only some basic requirements are needed, the main one being that road-load will be checkedonce the vehicle is in production. Within a certain tolerance this should comply to the road-load determined at type-approval.
These two approaches are schematically represented in Figure 1.
Figure 1 – Schematic representation of the feed-forward and feed-back approaches
For more information on these two approaches please refer to [Smeds 2011]. This will also list a number of known issues in the current road-load determination procedure that have caused the gap between road-load of homologation and production vehicles.
Outcome of the discussions over these approaches was that in any case the requirements and tolerances of ISO 10521 would need careful reviewing and where necessary be improved and/or tightened. This action is ongoing in the WLTP LabProcICE group, and has so far led to important improvements in the GTR draft text. The feed-back approach was however considered to be a ‘certification issue’ rather than a test procedure requirement, and should therefore be dealt with by the regional legislator that is responsible for implementing the new test procedure. The forum dealing with implementation of the WLTP in Europe is called the ‘EU-WLTP’ group, chaired by the EC. Also they are tasked with preparing the EU position towards WLTP issues and to ensure that the European conditions are reflected sufficiently in the cycle and the test procedure. This paper is not intended as a position paper but as an inputdocument for the discussions in EU-WLTP on the implementation of a feed-back approach for road-load determination in Europe, listing the possible options available and the details that need to be considered.
Introduction
Two of the important objectives of the WLTP are to create a level playing field for vehicle manufacturers and to increase the representativeness of the test cycle and measurement procedure. An improved road-load test procedure that reduces the gap between the road-load of homologation vehicles and production vehicles is expected to contribute in achieving both these objectives. There is considerable potential for improvement of the road-load test procedure, but it cannot be avoided that some roomhas to be given in the requirements and tolerances to allow for test-to-test and vehicle-to-vehicle variations.
It is recognised that there is a clear conflict of interest between the vehicle manufacturers and the legislator:
- Main goal of the legislator is that the measured road-load during type approval is representative for the actual real-life road-load of production vehicles. This is an important prerequisite for obtaining meaningful results from the emission tests, on which the legislator can build effective policy measures.
- Main goal of the vehicle manufacturers is to achieve the lowest road-load within the allowed room of the test requirements and tolerances. This will positively influence the measured CO2emissions and fuel efficiency, leading to a competitive advantage e.g. due to a lower CO2taxation. Secondary objective is that the results need to be repeatable, comparable and reproducible, preferably at low costs.
Due to this conflict of interest and the unavoidable room that needs to given in the test procedure, the measured road-load during type-approval is likelyto be lower than the actual road-load of production vehicles. The gap is further increased by the fact that road-load is determined at a prototype vehicle which is well-prepared for type-approval, but will not necessarily be completely identical to the vehicle that leaves the production line later on. Furthermore, this vehicle will be tested at a special test facility and under ideal test conditions, typically not representative for the everyday use of the vehicle.
The only way to ensure that these road-loads will generally correspond is by checking the actual road-load of production vehicles. This is the general idea on which the feed-back approach is based. There are some clear benefits to this approach:
- The road-load test requirements related to the vehicle specifications could be simplified since they are implicitly met once the road-load of production vehicles hasproven to comply.
- Not only the type approval, but all subsequent tests on the chassis dynamometer will use the correct road-load setting, including tests for Conformity of Production and In-Service Conformity requirements (please note that under the current NEDC procedure all laboratory tests, including those by independent test houses, are performed with unrealistic low road-load settings).
- It may be expected that in the near future on-road measurements with PEMS are introduced in legislation (refer to the developments in the RDE working group). Such measurements implicitly use the actual road-loadof the production vehicle, making the results better comparable to the type-approval test (even for the improved WLTP). It should be noted that RDE is focussing on off-cycle pollutant emissions, not on CO2.
At the same time there are some obvious drawbacks to the feedback-approach as well, such as the extra costs involved with checking the road-load on production vehicles, and selection criteria for a vehicle that is representative for the vehicle family. During the discussions in WLTP the manufacturers emphasised that road-load can vary from vehicle to vehicle, and is dependent on the way it is used by its owner. Such issues have to be addressed, possibly reflected e.g. in the allowed tolerance.
In this paper we will explore different options for implementation of the feed-back approach, while focusing on the issues of cost-effectiveness and practicability.
Options for feed-back approach
In general, there are two solutions identifiedto fulfil the ‘feed-back loop’ of Figure 1:
- Check on paper
Once the vehicle is type-approved and in production, it is verified if the production vehicle complies in all its road-load relevant components and settings to the specifications of the homologation vehicle. - Check by measurement
Once the vehicle is type-approved and in production, a run-in vehicle will be used to determine its road-load. The measured road-load is then compared with that of the homologation vehicle, and should comply within a reasonable tolerance.
Option 1: Check on paper
Actually, the check on paper could be seen as a ‘hybrid’ solution. It would need the same amount of requirements and tight tolerances as in the feed-forward approach, to be able to verify these at the production stage of the vehicle. The feed-back loop is not based on the actual road-load, but only verifies these requirements and tolerances. It is assumed that if those are compliant, the road-load is similar too. However, the flexibilities available in the test procedure and the offered room in the tolerances could still be exploited by the manufacturer, though be it to a lesser extent. There is no guarantee that the road-loads are identical. The level to which the road-loads of homologation and production vehicles will match strongly depends on the tightness of the requirements and tolerances.
On the positive side, it is an option that is probably very easy to implement. It would only require additional checking from the moment that vehicle production has started. The current ECE Regulation 83 already has provisions to physically check the ‘Conformity of Production’ (CoP), which would be an ideal placeholder to include a check on the road-load relevant components and factory settings. From that point of view it is regarded as a very cost-effective solution, as it involves no additional testing and yields only little extra administration burden on the manufacturer. Another advantage is that checking the specifications is exact and needs no extra tolerance: each component or specification is either compliant or not. One open issue that will need to be resolved concerns the consequences in case of non-conformity, e.g. a financial penalty, remedial measures, or a repetition of the type-approval. A suitable sanction needs to be considered together with the timing and frequency of the checking. Generally, a more severe sanction should be applied if the frequency of the checks is lower and/or the timing is later in the production stage. Such implementation details will be addressed later.
In summary, the check on paper has the following positive and negative attributes:
Easy to implement, e.g. in CoP.
Cost-effective solution, little extra administrative burden.
Simple verification procedure, without need for extra tolerances.
No guarantee that road-load of production vehicle is representative (expected to be higher)
Large number oftight requirements and tolerances needed in the test procedure.
Option 2: Check by measurement
As already mentioned in the introduction, the only way to make sure that the road-load of the homologation vehicle is representative is to verify the actual road-load of a production vehicle. A certain tolerance for the verification will have to be permitted, to allow for road-load variation between individual vehicles, test-to-test variation, etc.
Though it does not sound that complicated to verify the road-load of a production vehicle, it involves a number of practical difficulties. For instance, if a new vehicle is taken from the production line, it first needs to be run-in for at least 10,000 km. This is rather costly and time-consuming. Itmakes sense to let this run-in procedure be done by the manufacturer, but as a consequence of the fact that the vehicle offered for road-load verification is known, it might receive some special treatment.
A more objective verification is obtained if the road-load is checked on an in-use vehicle. However, the road-load will depend on how the vehicle is taken care ofby its owner, justifying a larger tolerance for the range within which the road-load is considered to comply. Furthermore, owners of the objected test vehicle have to be contacted and found cooperative to let their vehicle be tested by offering sufficient compensation and/or an alternative vehicle. Considerable costs will be involved to conduct the road-load verification on in-use vehicles.
Underthe current ECE Regulation 83, vehicle manufacturers in Europeare required to test the durability of the emission performance of their vehicles by conducting an In-Service Conformity test program. The tests needed for that program are done on in-use vehicles. The NEDC testcycle is tested on a chassis dynamometer to check whether the in-service emissions conform to the type-approval values. The setting of the chassis dynamometer is the same as that of the homologation vehicle, but it would make more sense if also the actual road-load would be used. The ISC requirement in the legislation could therefore serve as a placeholder for including the road-load verification. By using the actual road-load to set the chassis dynamometer a check on ISC of emissions becomes more realistic.[1]
Another alternative could be to let the checking of road-load be organised by the EU member states. Independent test houses could be tasked with performing a certain amount of verification tests on a random selection of in-use vehicles, similar to how in-use compliance programs for emissions are run in some member states[2].
Compared to a check on paper this option will deliver a more representative road-load, but due to the higher costs some compromises will be needed, e.g. a smaller sample size. If the check on road-load would be combined with ISC, road-load would only have to be checked on one or two vehicles of the family. The actual road-load could then also be used for checking emission conformity on in-use vehicles later on.
In summary, the check by measurement has the following positive and negative attributes:
Best guarantee on a representative road-load at type-approval.
Implementation possible as add-on to In-Service Conformity requirements or organised as independent in-use verification
Road-load determination procedure can be kept simple, with limited vehicle requirements.
Extra tolerance range needed to allow for production variance and possibly in-use influences.
Practical difficulties and relatively high costs.
Implementation details
There are a number of details that need careful consideration when a road-load checking procedure is implemented. Some of these have already briefly been touched, but for a complete overviewthey will be addressed here separately.Where applicable, possible benefits, drawbacks and/or trade-offs are identified, as input for the discussions between the stakeholders.
Sample size
In an ideal situation, road-load and emission performance would be checked on any individual vehicle on the road. Obviously that is not realistic, since there are high costs involved for this checking. Therefore, the sample size needs to be balanced against the extra costs involved. Secondly, there may be no need to check every vehicle when it can be proven that the road-loadis more or less consistent between identical vehicles. Though there is no data available to check the consistency of the road-load, it is expected that it should be able for manufacturers to control it within a certain range. In that case it would be sufficient to check only one vehicle of the family, assuming that the rest will comply as well if it complies within tolerance. If road-load verification would be organised outside the type approval legislation, the sample size could even be lower. As long as there are stringent sanctioning measures in place, it would be sufficient to test a small sample size of random selected vehicles (e.g. based on the sales numbers).
For the check on paper (option 1), the added costs are rather limited.It would be reasonable to demand that requirements and tolerances are confirmed on all vehicles that are checked within CoP.
Sample timing and frequency
The sooner a vehicle is tested after its production has started, the higher the potential impact of the road-load verification will be. If a non-compliant road-load is found towards the end of the vehicle’s production life, there is no effect on all the vehicles produced up to that moment of discovery. Therefore it is better to select the timing for verification close to the start of the production.
If verification is based on in-use vehicles, some time will be needed to have these vehicles run in properly. It is not considered relevant to check the in-use road-load of one vehicle family multiple times since conformity of the in-use road-load to the homologation road-load would principally only have to be demonstrated once. That is under the assumption that the road load remains more or less constant over the useful life of the vehicle (after the run-in period) and that there is no benefit to the manufacturer to introduce modifications to the production vehicle that would change the road-load characteristics. Currently there is no information to either support or reject these assumptions.