skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

This content will become publicly available on January 3, 2020

Title: Predicting Cycle-to-Cycle Variation With Concurrent Cycles in a Gasoline Direct Injected Engine With Large Eddy Simulations

Abstract

High cycle-to-cycle variation (CCV) is detrimental to engine performance, as it leads to poor combustion and high noise and vibration. In this work, CCV in a gasoline engine is studied using large eddy simulation (LES). The engine chosen as the basis of this work is a single-cylinder gasoline direct injection (GDI) research engine. Two stoichiometric part-load engine operating points (6 BMEP, 2000 RPM) were evaluated: a non-dilute (0% EGR) case and a dilute (18% EGR) case. The experimental data for both operating conditions had 500 cycles. The measured CCV in IMEP was 1.40% for the non-dilute case and 7.78% for the dilute case. To estimate CCV from simulation, perturbed concurrent cycles of engine simulations were compared to consecutively obtained engine cycles. The motivation behind this is that running consecutive cycles to estimate CCV is quite time-consuming. For example, running 100 consecutive cycles requires 2–3 months (on a typical cluster), however, by running concurrently one can potentially run all 100 cycles at the same time and reduce the overall turnaround time for 100 cycles to the time taken for a single cycle (2 days). The goal of this paper is to statistically determine if concurrent cycles, with a perturbation applied tomore » each individual cycle at the start, can be representative of consecutively obtained cycles and accurately estimate CCV. 100 cycles were run for each case to obtain statistically valid results. The concurrent cycles began at different timings before the combustion event, with the motivation to identify the closest time before spark to minimize the run time. Only a single combustion cycle was run for each concurrent case. The calculated standard deviation of peak pressure and coefficient of variance (COV) of indicated mean effective pressure (IMEP) were compared between the consecutive and concurrent methods to quantify CCV. It was found that the concurrent method could be used to predict CCV with either a velocity or numerical perturbation. Here, a large and small velocity perturbation were compared and both produced correct predictions, implying that the type of perturbation is not important to yield a valid realization. Starting the simulation too close to the combustion event, at intake valve close (IVC) or at spark timing, under-predicted the CCV. When concurrent simulations were initiated during or before the intake even, at start of injection (SOI) or earlier, distinct and valid realizations were obtained to accurately predict CCV for both operating points. By simulating CCV with concurrent cycles, the required wall clock time can be reduced from 2–3 months to 1–2 days. Additionally, the required core-hours can be reduced up to 41%, since only a portion of each cycle needs to be simulated.« less

Authors:
 [1];  [1];  [1];  [2];  [2];  [2]
  1. Convergent Science, Inc., Madison, WI (United States)
  2. Argonne National Lab. (ANL), Argonne, IL (United States)
Publication Date:
Research Org.:
Argonne National Lab. (ANL), Argonne, IL (United States)
Sponsoring Org.:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Vehicle Technologies Office (EE-3V)
OSTI Identifier:
1570971
Grant/Contract Number:  
AC02-06CH11357
Resource Type:
Accepted Manuscript
Journal Name:
Journal of Energy Resources Technology
Additional Journal Information:
Journal Volume: 142; Journal Issue: 4; Conference: Proceedings of the ASME 2018 Internal Combustion Engine Fall Technical Conference, San Diego, CA (United States), 4-7 Nov 2018; Journal ID: ISSN 0195-0738
Publisher:
ASME
Country of Publication:
United States
Language:
English
Subject:
42 ENGINEERING; energy conversion/systems; fuel combustion

Citation Formats

Probst, Daniel, Wijeyakulasuriya, Sameera, Pomraning, Eric, Kodavasal, Janardhan, Scarcelli, Riccardo, and Som, Sibendu. Predicting Cycle-to-Cycle Variation With Concurrent Cycles in a Gasoline Direct Injected Engine With Large Eddy Simulations. United States: N. p., 2019. Web. doi:10.1115/ICEF2018-9722.
Probst, Daniel, Wijeyakulasuriya, Sameera, Pomraning, Eric, Kodavasal, Janardhan, Scarcelli, Riccardo, & Som, Sibendu. Predicting Cycle-to-Cycle Variation With Concurrent Cycles in a Gasoline Direct Injected Engine With Large Eddy Simulations. United States. doi:10.1115/ICEF2018-9722.
Probst, Daniel, Wijeyakulasuriya, Sameera, Pomraning, Eric, Kodavasal, Janardhan, Scarcelli, Riccardo, and Som, Sibendu. Thu . "Predicting Cycle-to-Cycle Variation With Concurrent Cycles in a Gasoline Direct Injected Engine With Large Eddy Simulations". United States. doi:10.1115/ICEF2018-9722.
@article{osti_1570971,
title = {Predicting Cycle-to-Cycle Variation With Concurrent Cycles in a Gasoline Direct Injected Engine With Large Eddy Simulations},
author = {Probst, Daniel and Wijeyakulasuriya, Sameera and Pomraning, Eric and Kodavasal, Janardhan and Scarcelli, Riccardo and Som, Sibendu},
abstractNote = {High cycle-to-cycle variation (CCV) is detrimental to engine performance, as it leads to poor combustion and high noise and vibration. In this work, CCV in a gasoline engine is studied using large eddy simulation (LES). The engine chosen as the basis of this work is a single-cylinder gasoline direct injection (GDI) research engine. Two stoichiometric part-load engine operating points (6 BMEP, 2000 RPM) were evaluated: a non-dilute (0% EGR) case and a dilute (18% EGR) case. The experimental data for both operating conditions had 500 cycles. The measured CCV in IMEP was 1.40% for the non-dilute case and 7.78% for the dilute case. To estimate CCV from simulation, perturbed concurrent cycles of engine simulations were compared to consecutively obtained engine cycles. The motivation behind this is that running consecutive cycles to estimate CCV is quite time-consuming. For example, running 100 consecutive cycles requires 2–3 months (on a typical cluster), however, by running concurrently one can potentially run all 100 cycles at the same time and reduce the overall turnaround time for 100 cycles to the time taken for a single cycle (2 days). The goal of this paper is to statistically determine if concurrent cycles, with a perturbation applied to each individual cycle at the start, can be representative of consecutively obtained cycles and accurately estimate CCV. 100 cycles were run for each case to obtain statistically valid results. The concurrent cycles began at different timings before the combustion event, with the motivation to identify the closest time before spark to minimize the run time. Only a single combustion cycle was run for each concurrent case. The calculated standard deviation of peak pressure and coefficient of variance (COV) of indicated mean effective pressure (IMEP) were compared between the consecutive and concurrent methods to quantify CCV. It was found that the concurrent method could be used to predict CCV with either a velocity or numerical perturbation. Here, a large and small velocity perturbation were compared and both produced correct predictions, implying that the type of perturbation is not important to yield a valid realization. Starting the simulation too close to the combustion event, at intake valve close (IVC) or at spark timing, under-predicted the CCV. When concurrent simulations were initiated during or before the intake even, at start of injection (SOI) or earlier, distinct and valid realizations were obtained to accurately predict CCV for both operating points. By simulating CCV with concurrent cycles, the required wall clock time can be reduced from 2–3 months to 1–2 days. Additionally, the required core-hours can be reduced up to 41%, since only a portion of each cycle needs to be simulated.},
doi = {10.1115/ICEF2018-9722},
journal = {Journal of Energy Resources Technology},
number = 4,
volume = 142,
place = {United States},
year = {2019},
month = {1}
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on January 3, 2020
Publisher's Version of Record

Save / Share: