(a) General provisions. Use good laboratory practice to collect samples to represent the batch you are testing. For example, take steps to ensure that a batch is always well mixed before sampling. Also, always take steps to prevent sample contamination, such as completely flushing sampling taps and piping and pre-rinsing sample containers with the product being sampled. Follow the procedures in paragraph (b) of this section for manual sampling. Follow the procedures in paragraph (c) of this section for automatic sampling. Additional requirements for measuring RVP are specified in paragraph (d) of this section. A description of how to determine compliance based on single or multiple tests on single or multiple samples is specified in paragraph (e) of this section.

(b) Manual sampling. Perform manual sampling using one of the methods specified in ASTM D4057 (incorporated by reference; see § 1090.95) to demonstrate compliance with standards as follows:

(1) Collect a “running” or “all-levels” sample from the top of the tank. Drawing a sample from a standpipe is acceptable only if it is slotted or perforated to ensure that the drawn sample properly represents the whole batch.

(2) Use tap sampling (or other spot sampling) to collect upper, middle, and lower samples. Collect samples that most closely match the recommendations in ASTM D4057. Adjust spot sampling for partially filled tanks as shown in Table 1, Table 5, or Table 6 of ASTM D4057, as applicable.

(3) If the procedures in paragraphs (b)(1) and (2) of this section are impractical for a given storage configuration, you may use alternative sampling procedures as specified in ASTM D4057. This applies primarily for sampling with railcars, trucks, retail outlets, and other downstream locations.

(4) Test results with manual sampling are valid only after you demonstrate homogeneity as specified in § 1090.1337. Once a batch meets homogeneity specifications, you may use any properly drawn sample to represent the batch, subject to the hand-blending provisions of § 1090.1340. The entire batch volume is noncompliant if a sample fails to meet any applicable per-gallon standard.

(5) Except as specified for marine vessels in § 1090.1605, you must not do certification testing with a composite sample from manual sampling.

(c) Automatic sampling. Perform automatic sampling as specified in ASTM D4177 (incorporated by reference; see § 1090.95), with the additional provisions specified in this paragraph (c). Automatic sampling is intended to apply for in-line blending, including configurations that involve a pipeline filling a tank that will be certified as compliant before it leaves the fuel manufacturing facility gate.

(1) Follow all specifications for automatic sampling in this paragraph (c) unless EPA approves an in-line blending waiver that authorizes specific exceptions under § 1090.1315.

(2) Configure the system to ensure a well-mixed stream at the sampling point. Align the start and end of sampling with the start and end of creating the batch.

(3) Set a sampling frequency to represent a batch by meeting one or more of the following specifications, keeping records to show that the sampling frequency meets specifications:

(i) Collect 9,604 grabs.

(ii) Collect a number of grabs that achieves a margin of error of 0.01 or less as specified in Section 19.1.3 of ASTM D4177.

(iii) Collect grabs at regular intervals that do not exceed 20 seconds throughout the batch.

(4) Collect three samples for individual measurements in addition to the composite sample. Draw head, middle, and tail samples that come from the initial, middle, and final thirds of the estimated batch volume, respectively.

(5) If the composite sample fails to meet any applicable per-gallon standard, the entire batch volume is noncompliant. If one or more separate samples fail to meet any applicable per-gallon standard, the volume of noncompliant fuel is the volume starting with the last valid passing result before the failing result (or the start of the batch), up to the first valid passing result after the failing result (or the end of the batch).

(6) EPA may approve a different sampling strategy under an approved in-line blending waiver under § 1090.1315 if it is appropriate for a given facility or for a small-volume batch.

(d) Sampling provisions related to measuring RVP of summer gasoline. The following additional provisions apply for preparing samples to measure the RVP of summer gasoline:

(1) Meet the additional specifications for manual and automatic sampling in ASTM D5842 (incorporated by reference; see § 1090.95).

(2) If you measure other fuel parameters for a given sample in addition to RVP testing, always measure RVP first.

(e) Testing and reporting to demonstrate compliance with standards. Perform testing as specified in this subpart and report values to demonstrate compliance with per-gallon and average standards as follows:

(1) For parameters subject to per-gallon standards, report the highest measured value (or the lowest measured value for testing related to cetane index or other parameters that are subject to a standard representing a minimum value). This applies for repeat tests on a given sample and for testing multiple samples (including head, middle, and tail samples from automatic sampling).

(2) In the case of automatic sampling for parameters subject to average standards, report the result from the composite sample to represent the batch for demonstrating compliance with the average standard. For any repeat testing with the composite sample, calculate the arithmetic average from all tests to represent the batch.

(3) In the case of manual sampling for parameters subject to average standards, determine the value representing the batch as follows:

(i) For testing with only a single sample, report that value to represent the batch. If there are repeat tests with that sample, report the arithmetic average from all tests to represent the sample.

(ii) For testing with more than one sample, report the arithmetic average from all tested samples to represent the batch. If there are repeat tests for any sample, calculate the arithmetic average of those repeat tests to determine a single value to represent that sample before calculating the average value to represent the batch.