Reprinted from American Laboratory News August 1996

BY GLENN N. MERBERG

Evaluation of an octane analyzer

THE OCTANE NUMBER associated with a motor fuel is a measure of the performance of that fuel. The number does not correspond to the concentration of any one constituent in the fuel, but rather to the preignition properties of the fuels as compared with standard fuel mixtures. The term "octane number" is derived from the fact that the standard fuels contain isooctane.

The ASTM Standard Procedures for determination of octane numbers require the use of two Waukesha co-operative Fuel Research (CFR) engines (Waukesha Engine Div., Dressier Industries, Inc., Waukesha, WI). The Research Engine runs at 600 rpm, with a controlled air mixture and temperature as defined in ASTM D 2699. The Motor Engine runs at 900 rpm, with a controlled air mixture and temperature and controlled fuel temperature as defined in ASTM D 2700. These engines, or "knock engines," necessarily consume the fuel sample and a series of standard fuels during the test procedure.

The octane number of a fuel is defined by the compression at which the knock intensity of that fuel is 50% of the maximum knock intensity for the engine. Pure isooctane is defined as having an octane number of 100, while heptane is defined as having an octane number of 0. A series of standards containing heptane, isooctane, and toluene are used to define a scale between these two bounds. The research octane number (RON) of a fuel is the octane number measured on the Research Engine. Similarly, the motor octane number (MON) is the number determined by the Motor Engine. The octane number typically posted on the pump at a service station is the average of the RON and MON (called antiknock index [AKI], or [R + M]/2).

An octane analyzer

The ZX 101C (JSITS, Inc., Hagerstown, MD) is a portable, battery-powered octane analyzer for use with gasoline. It consists of three primary components: the analyzer, a sample container, and a light shield. The entire package in a carrying case weighs less than 5 kg. The instrument performs an octane number determination in less than I min and does not require the use of standard samples. The measurement is completely nondestructive.

Diagram of ZX101C optical systemThe analyzer measures octane number via near-infrared (NIR) transmission spectroscopy. The instrument contains a patented solid-state optical system comprising 14 near-infrared emitting diodes (IREDS) with narrow bandpass filters, a silicon detector system, and a fully integrated microprocessor. Figure 1 shows a schematic representation of the analyzer. The sample holder is a scaled, flat-sided, reusable glass container with an optical pathlength of 75 mm. The sample volume is approx. 225 mL.

To make an octane number determination, the user acquires a background signal from the empty sample chamber, measures the absorption spectrum of the sample twice, then acquires a second background signal. The entire process requires less than I min and can be performed by untrained, unskilled personnel.

NIR spectroscopy

The NIR spectral region is usually defined as that portion of the spectrum with wavelengths in the range 700-2500 nm. In this spectral region, overtone frequencies of molecular vibrations absorb light quite readily. Because the overtone absorption bands are typically wide and overlapping, spectroscopists cannot merely measure peak heights to perform quantitative analysis. Instead, multivariate regression analyses are utilized to correlate spectral features with concentrations or physical properties of interest.

The ZX101C operates in the short-wavelength NIR, from 800 to 1100 nm wavelength. The instrument is factory calibrated to predict octane number (RON, MON, and PON) from the absorption spectra of the fuels being tested. This prediction is accomplished through the use of a multivariate regression equation of the form:

Octane number = K0 + K1 (OD1) + K2 (OD2)...+ K14 (OD14) + K15 (Ta)

where K0 is a bias term, K1 through K15 are slope coefficients, OD1 through OD14 are the absorbencies measured at each of the 14 wavelengths, and Ta is the ambient temperature at the time of the test. The instrument can store up to 10 calibration equations and is factory calibrated for RON, MON, and PON of blended gasolines.

Analyzer features

The analyzer has several features that safeguard against erroneous readings: high variance warning - triggered if repeat measures for a given sample exceed some preset value; temperature warning - triggered if the analyzer is used outside of the recommended temperature range (15-45 degrees Celsius): out-of-range warning - triggered if the sample being tested is outside the calibration range of the analyzer, and curve checking - triggered if the optical spectrum of the current sample is atypical of gasoline samples stored in the unit.

Limitations imposed by CFR engine testing

Quantitative NIR spectroscopy is a secondary technique, in that results are generated through regression equations that model the output of some primary method of analysis. In the case of octane testing, the primary method of analysis is the CFR engine.

A secondary method can never show better accuracy than the precision of the primary method. This is true because accuracy cannot be determined beyond the bounds set by the primary method. In the case of CFR engine testing, the limitations are as summarized in Table 1. In this context, reproducibility describes the ability of an engine to generate the same result for a given' fuel after being shut down, then restarted and used by a new technician. Repeatability describes the ability of an engine to generate the same result for a fuel sample in successive tests. The 95% confidence interval corresponds to roughly two times the standard deviation of measurement for large numbers of retests.

ASTM specifications for CFR testing


As per procedure Reproducibility
(95% confidence)
Repeatability
(95% confidence)
Octane number
RON 2699 +/- 0.7 +/- 0.2 95
MON 2700 +/- 0.9 +/- 0.3 85

Table 1

Validation studies

State-sponsored octane screening

An increasing number of state governments are requiring filling stations to undergo octane testing. Stations caught selling fuels with octane numbers significantly different from the values posted on the pumps are fined and, in some cases, closed down.

In a traditional test, gasoline samples are collected from various filling stations and brought back to the laboratory for testing on CFR engines. The testing is necessarily time consuming and expensive. For this reason, many states have begun using the ZX 101C to perform on-site octane screening. The analyzer makes octane determinations in seconds and eliminates the tremendous expense of CFR testing.

During one recent survey, the state of Maryland Motor Fuel Testing Laboratory (Jessup, MD) collected 161 gasoline samples from various filling stations. The samples were each measured with the analyzer and then tested on both the Research and Motor CFR engines. Figures 2,3, and 4 contain scatterplots of the RON, MON, and PON of these samples as measured by the analyzer and CFR engines. A statistical summary of the data is presented in Table 2..

scatter plot 1scatter plot 2scatter plot 3

In an independent survey, the Commonwealth of Virginia Motor Fuel Laboratory (Richmond, VA) collected 163 gasoline samples. As with the Maryland study, the samples were brought to the state laboratory and tested with the analyzer and CFR engines. Figure 5 shows the calculated PON as measured with the analyzer and the CFR engines. A statistical summary of the data is presented in Table 2

scatter plot 4

The results of both of these validation studies show that the analyzer provides sufficient accuracy for the screening of gasoline samples from filling stations.

Octane exchange groups

Laboratories that make use of CFR engines typically belong to one or more octane exchange groups (OEGs). These groups circulate gasoline samples to the member laboratories and compile statistics about the octane numbers determined for similar samples by different laboratories. The purpose of the exchange groups is to allow laboratories to confirm that their engines are generating accurate data.

JSITS recently participated in one OEG's roundrobin test. A set of 20 samples was provided to the company with no laboratory data. The samples were measured on the analyzer, and the results returned to the OEG.

Later, the compiled laboratory values for RON and MON as determined by the OEG were returned to the company. Scatterplots for RON, MON, and PON, as determined by both the analyzer and the OEG, are shown in Figures 6, 7, and 8, respectively. A statistical summary of data is presented in Table 2.

scatter plot 5scatter plot 6scatter plot 7

Because the initial calibration in the analyzer was performed with one set of CFR engines; it is possible that the instrument may produce results that have a relative bias to other sets of CFR engines. For this reason, the analyzer is generally slope and bias corrected upon installation to provide results that are compatible with the customer's CFR engines. The OEG round-robin test did not afford JSITS the opportunity to perform slope and bias correction, because it was a blind test. However, after the test was complete and the analytical values for all the samples were known, slope and bias corrections were performed. The statistics presented for the corrected data reflect a best-fit regression performed between the CFR engine values and ZX 101C predicted values. The OEG data are provided on an as-is basis and with correction for slope and bias adjustment.

Summary Of validation studies
Test Site Test type Standard error
of estimate
Correlation
Maryland RON 0.55 0.97
MON 0.59 0.95
PON 0.46 0.98
Virginia PON 0.49 0.98
OEG RON 0.66 0.98
MON 0.47 0.98
PON 0.47 0.98
OEG
(slope and
bias corrected)
RON 0.45 0.99
MON 0.39 0.99
PON 0.30 0.99

Table 2

Precision studies

In order to assess the precision of the analyzer, the author obtained an instrument from the manufacturer and performed two experiments. In the first test, a single user tested a single sample 25 times in succession. The second experiment utilized five different technicians to test a single sample five times each. For each experiment, the mean, standard deviation, and range were calculated. Tables 3 and 4 summarize the results of these tests. The 95% confidence intervals are calculated using a Student's t distribution with 24 degrees of freedom.

Precision study: 25 retests by a single user
RON MON (R+M)/2
Mean value 98.4 87.9 93.2
95% confidence +/- 0.25 +/- 0.12 +/- 0.12
Maximum reading 98.6 88.1 93.3
Minimum reading 98.2 87.8 93.1

Table 3

Precision study: 25 retests
(five by each of five users)
RON MON (R+M)/2
Mean value 98.5 87.9 93.2
95% confidence +/- 0.25 +/- 0.12 +/- 0.10
Maximum reading 98.7 88.1 93.3
Minimum reading 98.1 87.8 93.1

Table 4

Summary

The analyzer was demonstrated to be an effective tool for the measurement of octane number in gasoline samples. Three independent studies provide validation of the instrument's utility. The instrument is extremely easy to use, portable, fast, and cost-effective. A typical analysis takes less than 1 min and requires no standard fuels and no experienced personnel. In general, the users seemed extremely satisfied with the analyzer and had already integrated the instrument into their regular octane testing program. Many commented that the analyzer provided cost savings by drastically reducing the number of CFR engine tests.

Reference
1. ASTM Procedures D 2699 and D 2700. 1995 Annual Books of ASTM Standards, Vol 5.02, 1995.

The data described herein were collected by independent laboratories, with the exception of the precision data, which were collected by the author.

Dr. Merberg is an Independent Consultant, 18004 Archwood Way, Olney, MD 20832. U.S.A.: tel.: 301-570-3246: email: gmerberg@aol.com. The author thanks Michael Jordan (Maryland Fuel Test Laboratory, Jessup, MD) and William Fowler (Commonwealth of Virginia Motor Fuels Laboratory, Richmond, VA) for sharing the results of their testing.
JSi@JSiTS.com

JSiTS