August 22, 1996 EPCD-SINR-96-02 Dear Under 19 kW Small Nonroad Engine Manufacturer: In this letter, the Environmental Protection Agency (EPA) will provide guidance on the calibration and span gas requirements described in the regulatory text in the Phase 1 new nonroad small SI engine rulemaking (60 FR 34582). Included in this document is guidance on determining compliance with calibration gas specifications, guidance on the best choice of zero and diluent gas types, and a discussion of a technical amendment to the Phase I rule to add a verification check on gas concentration when using a gas divider. A discussion of whether NIST Traceable Reference Materials (NTRMs) may be used in place of Standard Reference Materials (SRMs) is beyond the scope of the Phase 1 rulemaking and will be addressed separately. Calibration gas requirements are listed in section 90.312(c). One requirement of calibration gases is that the calibration gas values are to be derived from NIST SRMs and are to be single blends. Further, the regulations require the true concentration of a span gas to be within ± two percent of the NIST gas standard while the true concentration of a calibration gas must be within ± one percent of the NIST gas standard. The intent of the regulations is that gas standards be traceable to NIST gas standards (SRMs) and be within the stated levels of uncertainty with respect to NIST SRMs. For calibration gases the allowable uncertainty of ±1 percent typically means that the assigned value for calibration gas must be established, or verified, by direct comparison with NIST SRMs. For the manufacturer's records to be complete, the traceability and uncertainty of the calibration gas must be documented by the gas supplier on a certificate of analysis which is delivered with the gas. The certificate must identify the cylinder number, the gas mixture, the assigned concentration of the gas, the uncertainty of the assigned concentration, the SRMs used to certify the concentration, the date of the certification and the expiration date of the certification. Care must be taken in preserving the stability of calibration gases. For that reason, the expiration date of the certificate must be adhered to. Cylinder pressure also plays a part in the stability of gas standards. Some standards may become unstable when cylinder pressures drop below 500 psi. For span gases, the 2 percent uncertainty means that, at a minimum, their assigned value must be established or verified by direct comparison with calibration gases that meet the above criteria. This means that manufacturers may purchase a nominal ±5 percent standard and assign a concentration value to it by direct comparison with their calibration standards. In this situation, the manufacturer must take responsibility for documenting the traceability of the span gas. Some questions have been raised regarding the zero and diluent gases specified for the hydrocarbon analyzer's zero, span, and calibration gases. Section 90.312(c)(2) describes these requirements. This section allows the use of either purified synthetic air or purified nitrogen. There are several items to consider when deciding which gas to use. The following paragraphs give some background and recommendations in making this choice. Historically, the on-highway gasoline fueled engine test procedures used two different diluent gases. Purified synthetic air was used for dilute exhaust or CVS testing while purified nitrogen was used for raw emission testing. The reason for doing this was that the hydrocarbon analyzer -- the flame ionization detector or FID -- was sensitive to oxygen. By matching the oxygen level of the span/zero/calibration gas set to the oxygen level in the sample gas, the effect of oxygen is canceled out or at least minimized in the analysis of the sample gas. In raw gas testing, the sample gas contains a near zero level of oxygen, and thus the purified nitrogen was used as the diluent and zero gas. In CVS testing, the sample oxygen levels are closer to that of air which is why purified synthetic air was used as the diluent and zero gas. More recent experience suggests that newer hydrocarbon analyzers may be less sensitive to oxygen than were some of their older counterparts. In fact, the phase 1 rule test procedure does allow either diluent gas to be used with raw or CVS testing. Even so, it is still important to verify the hydrocarbon analyzer over a broad range of oxygen levels. For this reason, the phase 1 rule has a requirement to do an oxygen interference check to verify the fact that the hydrocarbon analyzer sensitivity to oxygen is minimized. The oxygen interference check is specified in 90.316(d). Consideration must also be given to high concentration mixtures of propane in air. Concentrations reaching the lean flammability limit are potentially unsafe and must not be used. In this situation, a mixture of propane and nitrogen should be substituted. Consult your gas supplier for guidelines on safe concentration levels for propane and air mixtures. Whichever diluent gas is chosen, it is still important to be consistent in the use of the zero gas and the diluent for the span and calibration gases. This will help to minimize variations from the hydrocarbon analyzer. For example, if purified nitrogen is chosen, then purified nitrogen should be the zero gas, and the diluent for the span gas and calibration gas. Section 90.312(c)(4) discusses some of the requirements applicable when using a gas divider. In addition to these requirements, EPA will be adding a check, adapted from section 86.114-94(a)(8) of the highway regulations, for manufacturers using gas dividers, to the Phase 1 technical amendments package. It will read as follows: "The use of precision blending devices (gas dividers) to obtain the required calibration is acceptable, provided that the calibration curves they produce name a calibration gas within 2 percent of its certified concentration. This verification shall be performed at between 15 and 50 percent of the full scale concentration of the range and shall be included with each gas calibration incorporating a blending device. Alternative procedures to verify the validity of the analyzer calibration curves generated using a gas divider are acceptable provided the procedures are approved in advance by the Administrator." Use of this check is not required until such time as the regulations are amended; however, EPA recommends that manufacturers begin to do this check for any future testing. When purchasing gas standards, manufacturers may realize some economy by specifying gas concentrations such that the standard used for the mid-span check on one range can be used as the calibration gas for a lower range. Please direct additional concerns and questions about calibration and span gas requirements to Anne Fredo, Certification Team Leader, at 202-233-9263. Sincerely, Robert F. Montgomery Group Leader Engines Compliance Programs Group (6403J)