The introduction of gasoline particulate filters (GPF) leads to new challenges in terms of calibration. FEV addresses these challenges with new development methods. The methods presented are simulation-supported RDE testing, use of specific worst-case fuels and accelerated ash loading.
The introduction of stricter particulate number (PN) limits in Europe and China is complemented by the measurement of Real Driving Emissions (RDE) on public roads at arbitrary boundary conditions. This creates a challenge to meet the PN limits and a clear trend for all OEMs to apply GPF systems in order to comply with current and future RDE legislations.
Up until now, automotive manufacturers have been able to meet the EU6d-TEMP PN limits for standardized driving cycles, such as the NEDC with the Worldwide Harmonized Light-Duty Vehicles Test Procedure (WLTP) without having to depend on a GPF. The left side of Figure 1 supports this statement, showing PN emissions of approximately 3 x 1011 1/km without a GPF for a J-class vehicle with a turbo DI gasoline engine in a WLTC test. The implementation of a GPF with an efficiency of around 75 percent leads to a further significant PN reduction.
The right hand side of Figure 1 shows influences that are valid under RDE conditions, which can lead to substantially higher particulate emissions compared to nominal WLTP conditions. The first factor shown, is the decrease in the share of the ethanol content in the fuel, e.g. from 10 vol-percent (EU6 fuel respectively “nom. Cond.”) to 0 vol-percent (EU6 worst-case fuel), which is caused by the reduction of the fuel´s oxygen share. Secondly, and also a fuel-related topic, a strong knock resistance of high aromatic content comes with a high particulate formation potential. A payload increase, caused by luggage and/or passengers, leads to engine operation with higher speeds and loads, which further increases the PN output (dark blue bar). Moreover, a sporty or aggressive driving style causes increased particulate numbers, especially when combined with the other illustrated influences. 
>> THE FEV WORST-CASE RDE SIMULATION ENABLES AN EVALUATION OF THE FULL RANGE OF BOUNDARY CONDITIONS
Considering these multiple aspects and all their possible combinations, the extensive integration of GPFs will become mandatory. Consequently, a need to develop suitable tools and methods becomes important, which enables the consideration of all GPF impacts and interactions in an early development phase. The following article discusses the simulation-supported RDE testing and highlights the current GPF hardware trends and calibration subtasks for ultra-low temperatures. It concludes with results of an accelerated GPF ash loading procedure.
Simulation-Supported RDE Testing
Often, the investigation of all influencing boundary conditions for the determination of RDE values requires consideration of several cycles and driving scenarios. If a large number of different cycles need to be tested, this leads to cost and time issues. Figure 2 shows the main RDE boundary conditions and a comparison of the WLTC chassis dyno cycle with an on-road RDE test (conducted on the FEV Aachen RDE track). In addition, a simulation-supported, tailor-made worst-case RDE driving scenario, conducted on a chassis dyno is depicted. The on-road Aachen RDE test is compliant with the law and covers sections in an average mountain range. The best-case condition of each criterion is marked by the center of the diagram, whereas the worst-case condition is marked by the outer border.
As seen in Figure 2, the WLTP covers quite severe speed conditions, and medium vapos and vehicle weight values. The vapos number describes the positive acceleration multiplied by the vehicle speed and is a first characteristic number to evaluate RDE driving. However, the influences of full load acceleration, cold start and positive altitude gain or temperatures are not covered completely. The RDE-compliant FEV Aachen RDE track covers significantly more of the diagram. But, only the FEV worst-case RDE simulation enables an evaluation of the full range of boundary conditions. It is clear that only a holistic simulation approach is able to cover all of these boundary conditions in one cycle.
>> A HIGH NUMBER OF PARAMETERS REQUIRES SIMULATION-BASED TESTS
GPF Hardware and Calibration Overview
Another fact supporting the trend of simulation-supported testing is the wide variety of GPF applications on the market. Figure 3 displays current technology and market trends for GPF applications based on FEV’s in-house analysis. Despite the fact that 75 percent of actual GPFs are coated, FEV expects a long-term trend of uncoated GPFs due to their backpressure advantages . Besides, most GPFs are located in close-coupled position in order to utilize the high exhaust gas temperatures for soot regeneration.
The integration of GPFs into a gasoline powertrain brings a number of additional calibration tasks. The major tasks are soot model calibration, monitoring of soot loading (simulation and DP sensor) and safety function calibration. All calibration tasks aim at minimizing the customer impact of the GPF implementation. Typical ECU calibration tasks are listed in Figure 4.
The most important input for GPF calibration is an engine-out soot map. For that, the base engine calibration must be in a mature state, including optimized particulate emissions. The soot map is usually based on worst-case fuel, as these conditions serve as a main driver for the initiation of active regenerations. Tests for GPF loading and (active) regeneration take place at the engine test bench. It is necessary to determine the critical specific soot mass for the filter to prevent thermal shock during regeneration. Furthermore, the corresponding tests generate input data for soot loading and oxidation models, as well as for the backpressure model. With the definition of the critical specific soot mass, the calibration of GPF monitoring and safety functions are possible. Both are of high importance for the impact on drivers. 
GPF and Cold-Start Conditions
As the worst-case fuel plays a central role within GPF calibration, FEV applied their fundamental fuel research experience regarding the impacts of ethanol or aromatics content in the fuel to a complying fuel. Since there is no fuel quality sensor in current series applications, the ECU must always consider worst-case fuel.
Figure 5 displays the GPF soot load after ten repeated cold starts at different temperatures. For EU6 certification, fuel with 10 vol-percent ethanol and the chosen SUV with a turbo DI engine, ten cold starts at -20 °C result in a soot load of approximately 0.9 g/l. For the same conditions, the FEV worst-case fuel produces about 50 percent more soot. A reference fuel shows the same behavior.
Ten repeated cold starts at -30 °C led to a specific soot load of 2 g/l, which is the threshold to trigger an active regeneration for the specific application. Considering two cold starts per day, a temperature of -30 °C may become critical regarding the GPF soot loading after only five days of client operation. Thus, it is mandatory to implement an ECU function for active filter regeneration.
Accelerated Ash Loading
In order to consider the aging effects during calibration, accelerated GPF ashing is carried out on the burner test bench. Figure 6 illustrates two oil-based fuel doping experiments on the burner test bench, However, since this aging method is known to produce ash with very high backpressure, the burner hardware has been modified to precisely control the ash properties. The modified hardware can optimize the ash formation and significantly reduce the backpressure, which leads to ash properties comparable to vehicle ash. This improvement was achieved without limiting the shortening factor and full utilization of the potential is still in development.
Thus, the burner ash generation is a cost- and time-effective tool for end-of-lifetime investigations with respect to the wide range in which field ash varies for different customer applications. Especially in early development stages, where no durability runs are finished, the burner aging enables GPF aging-effect calibration.
Due to the future measurements of real driving emissions, an entire new range of influencing factors of vehicle calibration must be considered in order to be certification compliant. This makes a GPF application mandatory. Currently, test cycles do not include RDE worst-case conditions. FEV developed a simulation tool that generates worst-case cycles in order to develop calibrations that guarantee RDE compliance under all boundary conditions.
An FEV market analysis confirms the increasing trend to GPF applications, but also shows that there are a variety of technologies and installations for different vehicle applications available. As the GPF affects the engine operation, new calibration tasks arise. In order to minimize the related calibration effort, the simulation-supported testing is combined with worst-case fuels and cold start conditions. As a result, the soot modelling and regeneration calibration tends to be on the safe side. Accelerated ash loading on the burner test bench addresses the evaluation of aging impacts and allows end-of-life GPFs at a very early development stage.