Daniel C. Zebelean
Northwest Pipeline Corp.
Salt Lake City
Northwest Pipeline Corp., Salt Lake City, has developed and employed monthly fuel-consumption reports to track differences between manufacturers' predicted fuel consumption and actual operating fuel consumption.
The purpose of the report system is to locate sources of fuel inefficiency in compressor-station operations. Initial results have indicated a potential savings of $128,000/year.
Northwest Pipeline Corp. (NWPL), a subsidiary of the Williams Cos., Tulsa, operates 7,000 miles of natural-gas pipeline with main line delivery capacity exceeding 2 bcfd serving nine western U.S. states.
With the onset of deregulating the natural gas industry, market competitiveness will be the primary factor in determining future survival.
One of the factors in determining Northwest Pipeline's cost of service is cost of engine fuel to transport natural gas. Last year, NWPL consumed approximately 9.6 quadrillion BTUs in engine fuel gas.
Although fuel gas has always been measured and accounted for, the thermal efficiency of the fuel gas consumed has never been quantified. In 1989, NWPL's engineering management set a goal to audit the fuel consumption monthly, comparing actual fuel consumed to manufacturer's predicted fuel-consumption curves.
The difference between actual fuel consumption and manufacturers' predicted fuel consumption was decreased from 7% to approximately 3% within 6 months after the first report was generated (Fig. 1).
Unbalanced engines, faulty spark plugs, fuel valves, engines requiring overhauls, and even computer software and hardware problems were discovered as a result of this initial report.
The total decrease in the comparison of 4% was not entirely realized. Approximately 1.5% of the decrease resulted from revisions in the manufacturers' fuel-consumption curves correctly to predict NWPL's vintage of engines.
[The manufacturers' fuel consumption curves are stated in heat rate (BTU/bhp-hr) not fuel consumption (Mcfd). The two terms are frequently interchanged, but by definition, are different. For simplicity, heat rate is referred to as fuel consumption in this article.]
After the manufacturers' curves were corrected, the average monthly heat rate for NWPL's system was 7,957 BTU/bhp-hr. The current monthly average heat rate is 7,752 BTU/bhp-hr.
A net decrease of 205 BTU/bhp-hr was achieved as a result of the fuel consumption report. In monetary terms, this equates to a savings of approximately $128,000/year.
(This estimate assumes operating 50,000 bhp-one third of NWPL's main line horsepower-365 days/year, fuel gas 1,050 BTU/cu ft, $1.50/Mcf.)
PREDICTING FUEL CONSUMPTION
NWPL's supervisory control and data acquisition (scada) system logs the necessary operating data required to extract the predicted fuel consumption from the manufacturers' fuel-consumption curves.
Hourly values of engine torque, rpm, and other operating parameters are transferred from the scada system to NWPL's IBM mainframe for access by any user in a relational data base called DB2.
With known engine torque and rpm, the predicted fuel consumption can be determined from the manufacturers' curves.
Fig. 2 is a generic fuel consumption graph for a reciprocating engine.
Normally, the graph has two curves for the predicted fuel consumption: one for 100% torque and varying speed, the other for 100% speed and varying torque.
As can be seen, at 100% speed and 100% torque, the fuel consumption would be 7,000 BTU/bhp-hr. At 100% speed and 70% torque, the fuel consumption would be 7,550 BTU/bhp-hr.
A third-degree polynomial equation was written for each predicted fuel-consumption graph. These equations are stored in a program, for each type of reciprocating engine.
Utilizing the scada system readings for torque and rpm, these equations determine the predicted fuel consumption.
The manufacturers' fuel consumption curves that NWPL utilizes are for the sea-level rating of the engines. Most of NWPL's engines are significantly above sea level and are derated to compensate for the decreased air density.
Therefore, an adjustment is necessary to predict the fuel consumption accurately when the engine is derated for altitude.
The scada system calculates 100% engine torque at the site-rated horsepower and 100% speed. The scadacalculated torque is adjusted to equate to the manufacturers' sea-level predicted, fuel consumption curves by multiplying the scada-computed torque by the ratio of the site-rated load to the sea level-rated load.
As an example, a reciprocating engine has a sea-level rating of 2,000 bhp and is operating at an altitude of 6,000 ft above sea level.
The engine would be derated to approximately 1,700 bhp and rated speed. It would be operating at 100% of the site-rated torque, which equates to 85% (1,700 bhp/2,000 bhp) of the sea level-rated torque.
Fig. 3 illustrates this concept.
At 100% sea level-rated torque and 100% speed, the predicted fuel consumption is 7,000 BTU/bhp-hr. At 100% site-rated torque and 100% speed, the predicted fuel consumption is 7,100 BTU/bhp-hr.
The difference in the sea level and site-rated torques' predicted fuel consumption is more significant at the lower end of speeds and torques.
AT REDUCED SPEED, TORQUE
The need for predicting fuel consumption at reduced speeds and torques has been minimal. Therefore, the engine manufacturers have not researched or published engine operating data predicting the fuel consumption at reduced speeds and reduced torques.
NWPL's approach assumes that the predicted fuel consumption at reduced speeds and torques is between the 1 00% speed, varying the torque curve, and the 100% torque, varying the speed curve.
Mathematical equations were written to interpolate data between these two curves to predict the fuel consumption when the engine is operating at reduced speeds and reduced torques. Fig. 4 illustrates these mathematical equations.
With the graph in Fig. 4 and assuming an engine operating at 90% torque and 80% speed, the predicted fuel consumption is 7,333 BTU/bhp-hr.
The percent torque and percent speed are multiplied to obtain percent load: 90% x 80% = 72%.
The predicted fuel consumption from the constanttorque curve (7,200 BTU/bhp-hr) and the constant speed (7,600 BTU/bhp-hr) are obtained by running the polynomial-stored equations at 72%.
The constant-torque fuel consumption and constantspeed fuel consumption from the polynomial equations are multiplied by ratios calculated from a weighted average of percent torque (90%) to percent speed (80%).
These numbers are then added together to yield the total predicted fuel consumption rate. (See accompanying equations box.)
In Equation 1, the difference between 100% speed and operating speed and the difference between 100% torque and operating torque are added together to obtain the constant Z.
In Equation 2, the ratios X and Y are determined from the weighted average divided by the constant Z.
As shown on the graph of Fig. 4, at 72% load:
(100 - 80) + (100 - 90) = 30 Fig. 4
X = (100 - 90)/30 = 0.3333; Y = (100 - 80)/30 = 0.6667
7,600 x 0.3333 + 7,200 x 0.6667 = 7,333 BTU/bhp-hr
In essence, this weighted average is stating that the 100%-speed fuel consumption will be multiplied by one third because the percent operating speed (80%) was twice as far from the rated 1 00% parameters as the percent operating torque (90%).
This method of predicting fuel consumption at reduced speeds and reduced torques tracks the actual fuel consumption within 4% at reduced speeds and reduced torques. Therefore, NWPL accepted as accurate this method of predicting the fuel consumption at reduced speeds and reduced torque.
CONSUMPTION MEASUREMENT
NWPL's scada system calculates and logs the fuel consumption of each unit in the DB2 data base.
An individual meter run for each engine measures the fuel flow. The AGA Report No. 3 (AGA3) calculation is used to derive total fuel flow to the engine.
The AGA3 equation uses several factors to calculate fuel flow. Differential pressure, static pressure, and specific gravity factors are measured variables, and the rest of AGA3 factors are assumed to be constant. The fuel consumption is stored in Mcfd.
The fuel consumption is converted from Mcfd to BTU/bhp-hr by multiplying by the BTU higher heating value (HHV) of the gas, dividing by the measured load on the engine, and multiplying by a constant of 37.5.
BTU/bhp-hr = ([Mcfd x HHV]/bhp) x 37.5
where:
37.5 = (Mcfd x 0.9 HHV/ HHV)/24-hr/day LHV = Lower heating value of the gas
The BTU HHV value of the gas and the engine load are measured and stored hourly in the DB2 data base. The BTU HHV value is measured by an on-line chromatograph. The engine load is measured by Kistler pressure transducers.
A Kistler transducer is a piezoelectric transducer that measures a pressure change. One is located in the head end and in crank end of each compression cylinder on the engine.
A pressure-volume curve is generated from each Kistler during the compression cycle (Fig. 5). The area inside the pressure-volume curve is calculated to measure work or load.
The load calculated from all of the pressure-volume curves from each head end and crank end are added together to yield the total load on the engine.
The scada percent engine torque is calculated by taking the total load calculated from the Kistler pressure transducers, dividing by the site-level rating of the engine, and multiplying by ratio of the rated speed-operating speed, as follows:
% torque = (calculated load/site rated load) x (rated speed/operating speed)
FUEL-CONSUMPTION REPORT
The fuel-consumption report consists of two sections shown in Figs. 6 and 7 The first section is a management summary report, and the second section consists of engine operating data extracted from the DB2 data base.
The management summary report condenses the data report to give management the pertinent information required to analyze engine-fuel consumption.
The report lists average engine torque, average total bhp, number of hourly points used for the averaging, the average thermal efficiency, the average fuel-consumption rate, and the average deviation of actual fuel consumption vs. predicted fuel consumption.
The management summary report flags ("$") an engine when the actual fuel consumption is 10% more than the predicted fuel consumption. An additional flag is displayed for each 10% increment, up to 50%.
For example, if the actual fuel consumption is 34% more than the predicted fuel consumption, then three flags will be displayed.
The summary report will also flag thermal efficiencies of less than 25% and greater than 40% and fuel consumption rates less than 6,300 BTU/bhp-hr and more than 12,000 BTU/bhp-hr. Flagging these abnormal thermal efficiencies and fuel consumption rates will indicate either erroneous data or an inefficient engine.
The second section of the fuel report consists of engine operating data.
A program is run to query the data base to select the engine operating data. Limitations are used in the program to select credible operating data. These limits are listed on the bottom of the management summary report (Fig. 6).
With these limits, erroneous data are minimized, and data are gathered within the predictable limits of the manufacturers' fuel consumption curves.
The fuel-consumption report has flagged several engines that were operating at more than 10% over the predicted fuel consumption. Deficient fuel valves, fowled spark plugs, bad transmitters, incorrect software factors, wrong top-dead-center offsets, and numerous other mechanical problems were pinpointed and corrected based on the fuel report.
Below are two actual examples of problems flagged by the fuel report.
In one instance, a 4,000-bhp engine was flagged for operating more than 30% over the predicted fuel consumption.
The engine, found to be operating out of balance, was balanced and the actual fuel consumption dropped to 5% over the predicted fuel consumption. Fuel savings of $8,000/month were realized.
Another case involved one station with four 2,000-bhp engines consistently operating 10-15% in excess of the predicted fuel consumption. The engine was in excellent operating condition but was difficult to balance properly.
After months of checking for the problem, the PFM 2000 engine analyzer was utilized to help solve the problem. A software factor in calculating engine load was found to be incorrect.
The engine was actually developing 10% more horsepower than the computer calculated. This miscalculation led to engine overload and detonation. The software factor was corrected, the detonation ceased, and actual fuel consumption dropped to 2% over the predicted fuel consumption.
The actual and predicted fuel consumption is now displayed on the local compressor station's CRT screen. This allows the station operators to compare the actual and predicted fuel consumption daily and repair any problems immediately if required.
The environment would benefit by fewer air-emission pollutants by the continuous monitoring of the engine operating condition. The fuel report flags inefficient engines caused by improper air-fuel or ignition control settings which result in high NOx or CO2 pollutants.
CUSTOMER BENEFITS
With the continuous monitoring of the specific fuel consumption, inefficient engines are flagged and corrected immediately, saving fuel cost and possibly costly engine mechanical failures. These savings are passed on to NWPL's customers, lowering the company's cost of service and increasing its competitive edge.
The primary objective of the fuel-consumption report was to compare the actual fuel consumption to the predicted fuel consumption to quantify NWPL's engine efficiencies. This objective was met. The results also indicate if there is a computer software or hardware problem, or if any of the engine-mounted hardware devices that measure operating data are failing.
These hardware devices are the same that control the engine loading, the air-fuel ratio control, and the ignition control. The fuel report, therefore, adds extra insurance that the engine automated controls are working properly.
With the modern technology of gathering, storing, and manipulating engine operating data, programs can be written to evaluate in a matter of seconds enormous amounts of engine operating data. This data output can then be formatted in concise reporting that is useful in evaluating operating efficiencies, such as the fuel report.
NWPL is in the process of writing other programs which use the operating data base as a source. The objective of these programs will be to increase operating efficiencies and lower maintenance cost, which ultimately lowers the cost of service to the customers.
ACKNOWLEDGMENTS
The author wishes to thank Dave Dean, John McAughan, and Steve McGee, all of Northwest Pipeline Corp., for their roles in developing the technology reported in this article.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.
