Optimizing blend targets requires balancing product quality compliance with the economic cost of unnecessary quality giveaway. Conservative targets can protect against measurement and execution uncertainty, but when they no longer reflect actual blend-control capability, they can increase blend costs and leave significant refinery margin unrealized. This article examines a structured approach to blend target optimization that combines the statistical compliance principles of ASTM D3244 with refinery-specific blend-control variability. Using RVP as a practical example, it considers how measurement uncertainty, process variability, operational risk and refinery-specific performance can be incorporated into statistically defensible targets. It also explores the role of backcasting, blend-law recalibration and bias management in continuously improving prediction accuracy and enabling targets to move closer to specification without unnecessarily increasing compliance risk. The result is a target-setting framework designed to reduce quality giveaway, decrease blend costs and capture additional refinery margin from existing operations.

One of the largest sources of value leakage in a refinery, often amounting to tens of millions of dollars annually, is the use of overly conservative blend targets that result in unnecessary product quality giveaway. Establishing optimized blend targets enables refiners to maximize margins by balancing product quality compliance risk against the cost of quality giveaway. By integrating the compliance framework defined in ASTM D3244 with refinery-specific blend control variability, consisting of execution variability, measurement uncertainty, and analyzer precision, blend targets can be optimized to consistently satisfy specifications while minimizing blend costs and reducing reliance on high-value blend components.

Refiners often maintain conservative targets for operational reasons. During blend scheduling and execution, uncertainty arises from blend control variability which is influenced by laboratory measurement error, analyzer precision, blend model performance, and variability in component quality. To protect against these uncertainties, organizations establish safety buffers in their target planning process. However, these buffers are frequently based on historical assumptions and may not accurately reflect current blend control capability. As a result, refiners can unknowingly operate further away from specification than necessary, increasing blend costs without decreasing compliance risk.

Because the point of compliance resides with the receiver, specification compliance is further complicated by the inherent variability of testing methods as the same sample tested by different laboratories may produce different results due to method reproducibility. ASTM D3244 was developed to provide a statistically defensible framework for interpreting specification compliance in the presence of method reproducibility. Rather than evaluating compliance solely against the published specification value, the standard defines acceptance limits that account for test method reproducibility and establish an appropriate balance between supplier and receiver risk.

As every test method exhibits inherent variability, quantified by its repeatability and reproducibility, a reported test result cannot be assumed to represent the true value of the property being measured. Instead, ASTM D3244 establishes acceptance limits that incorporate method reproducibility to manage the risk of incorrect compliance decisions and provide a consistent basis for product certification.

Within this framework, two reproducibility-based adjustment factors are commonly used. The 0.593R factor is applied when compliance is determined from a single test result and represents the standard ASTM D3244 acceptance limit corresponding to approximately a 95% confidence level, or about a 5% retest risk, and is typically used for regulatory, contractual, and other critical specifications. The 0.419R factor is applied when acceptance decisions are based on the average of two independent test results rather than a single reported result. Because averaging reduces measurement uncertainty, the resulting acceptance limit can be established closer to the specification limit while maintaining the same overall decision risk. Consequently, the 0.419R criterion is appropriate only when the average of two independent test results is used as the basis for compliance determination.

Refiners typically build on the ASTM D3244 acceptance-limit framework by incorporating additional operational targets that account for process variability. The release limit represents the minimum quality required for product disposition and is generally set as the contractual or regulatory specification itself. The acceptance limit is then calculated by applying the appropriate D3244 reproducibility adjustment to account for test-method uncertainty. Finally, an offset limit is often applied to address refinery-specific blend-control variability, resulting in the operational blend target used during production. In this hierarchy, the release limit protects compliance, the acceptance limit manages measurement uncertainty, and the offset limit manages process uncertainty. Together, these limits provide a structured approach for balancing compliance risk against quality giveaway.

The following examples illustrate how this framework can be applied to Reid Vapor Pressure (RVP) target setting, showing how acceptance limits and blend-control variability combine to establish economically optimized operating targets for both critical and non-critical specifications.

Summer Reid Vapor Pressure (RVP) provides a practical use case of target setting using D3244 methodology. Due to stringent EPA regulations, product quality groups will classify summer RVP as a critical specification requiring a high degree of confidence that delivered gasoline will meet regulatory limits.

For a ceiling specification, ASTM D3244 establishes an acceptance limit equal to:
Acceptance Limit = Specification − 0.593R
where R represents the reproducibility of the test method.

For gasoline with a 9.0 psi RVP specification measured using ASTM D5191, the method reproducibility is approximately 0.23 psi, resulting in an acceptance limit of:
8.86 psi = 9.00 − (0.593 × 0.23)

While releasing product at the acceptance limit statistically represents a 5% risk that a receiver’s test result exceeds specification, most refiners apply an additional safeguard known as an offset limit to further reduce the probability of a receiver obtaining an off-spec result. The offset limit is calculated as:
Offset Limit = 1.645 × Standard Deviation of Blend Control Variability

Assuming a standard deviation of blend control variability of 0.05 psi, the economically optimized summer RVP target becomes:
8.78 psi = 8.86 − (1.645 × 0.05)

This target accounts for both testing reproducibility and blend execution variability, effectively reducing the probability of a receiver measuring the product above the 9.0 psi specification while maintaining a structured, statistically defensible compliance approach.

In contrast, winter RVP is often considered a non-critical specification because regulatory constraints are less restrictive. In these cases, product quality groups may place less emphasis on ASTM D3244 acceptance limits and instead focus primarily on blend execution capability.

Using a winter specification of 13.5 psi and the same standard deviation of blend control variability of 0.05 psi, the target is calculated as:
13.42 psi = 13.50 − (1.645 × 0.05)

Because compliance risk is lower, the blend target can be set much closer to specification, reducing unnecessary giveaway while still maintaining operational confidence.

The effectiveness of any target-setting methodology ultimately depends on blend control variability, which measures the refinery’s ability to execute blend plans as intended. Blend control variability is typically quantified as the difference between predicted properties generated during recipe generation and the final certified product properties.

Unlike test method reproducibility, which is largely external to the refinery, blend control variability is directly controllable and therefore represents one of the greatest opportunities for margin improvement.

This variability is influenced by several factors, including:

• Blend model and blend law prediction accuracy
• Component property data quality
• Sampling variability
• Real-time blend execution control
• Online analyzer performance and reliability
• Measurement system precision and bias
• Certification methodology (online vs. laboratory measurements)

Refineries seeking to reduce giveaway must therefore focus on continuously improving each of these elements. As blend control variability improves, targets can be moved closer to specification, directly translating into lower blend costs and improved margin capture.

While ASTM D3244 and blend control variability form the foundation of an optimized target-setting framework, several practical considerations must also be incorporated.

Blending configuration is one such consideration. Inline blending often requires additional conservatism because all barrels blended into the pipeline need to be on-spec. Additionally, correcting an off-spec pipeline blend is significantly more challenging than correcting a blend going to a tank.

Receiver retest frequency is another important factor. ASTM D3244 implicitly assumes every batch may be retested by the receiving party; however, this assumption is not always representative of real-world practices. For example, octane properties are generally retested less frequently than RVP or distillation properties because ASTM D2699 and D2700 knock-engine testing is less accessible. Consequently, refiners may choose to operate closer to octane specifications while maintaining more conservative targets on properties that are routinely verified by customers.

Finally, organizational risk tolerance plays a critical role. The ASTM D3244 framework is based on a 5% probability of off-spec results; however, individual refiners may adopt either more conservative or more aggressive risk thresholds depending on their commercial strategy, regulatory environment, customer relationships, and historical product quality performance.

Blend target optimization should not be viewed as a one-time exercise. Continuous comparison of predicted blend properties against certified product results enables refiners to quantify blend control variability, identify systematic model bias, and measure the effectiveness of target-setting strategies. Through routine backcasting analyses, refiners can determine whether planning buffers remain appropriate or whether targets can be moved closer to specification without increasing compliance risk. This ongoing feedback loop helps ensure that target-setting practices remain aligned with current refinery performance and changing operating conditions.

As an example, RVP blend laws should be reviewed and recalibrated before each blending season using historical data from the corresponding season, such as using the previous summer’s blend data to establish exponents for the upcoming summer season. Backcasting analyses can then be used to develop site- and grade-specific exponents that more accurately represent the nonlinear relationship between butane concentration and blended RVP. This is important because RVP blend behavior varies with gasoline composition, grade, and seasonal butane utilization. Premium grades often exhibit different blending characteristics than regular grades, while higher winter butane concentrations typically result in greater nonlinearity than summer blends. Consequently, a single static exponent is unlikely to accurately represent the full range of refinery operating conditions.

The preferred approach is to use periodic backcasting and model recalibration to maintain seasonal- and grade-specific blend-law exponents that capture the underlying long-term blending behavior. Once the structural blend law has been established, bias management can be applied as a secondary control layer to compensate for short-term influences such as analyzer bias, process drift, changing component qualities, or shifts in operating conditions. In this framework, blend-law revisions address structural prediction error, while bias management corrects temporary deviations from expected performance. Together, these practices create a continuous improvement cycle that improves prediction accuracy, reduces blend-control variability, and enables targets to be set closer to specification.

Combining the statistical compliance principles of ASTM D3244 with refinery-specific blend control performance provides a structured and economically optimized approach to blend target setting. This methodology minimizes unnecessary quality giveaway, reduces blend costs, improves utilization of lower-value components, and maximizes refinery margin while maintaining product quality compliance.

When coupled with disciplined blend reviews, backcasting analyses, and sustainability processes that continuously measure and improve blend control variability, the framework becomes a powerful mechanism for sustained performance improvement. Refineries that establish formal target-setting governance and align execution to economically optimized operating targets can unlock significant value and create a lasting competitive advantage.