Pressure Influences Power

By Mark Krisa


With current financial and market pressures, energy conservation has become a priority. Compressed air is a
recognized energy consumer with great opportunities for energy savings with a quick payback. Reducing air system pressure is a common action to drive energy savings but it is typically not well received by production and the potential savings can be misleading. In an effort to drive results, this savings opportunity is frequently over simplified and the savings potential exaggerated. Greater detail regarding this action is discussed in hopes of clarifying issues and improving calculated returns.


Energy conservation measures associated with compressed air have received a significant amount of attention over the years, mostly due to a reasonably short financial return compared with other energy consuming equipment. Over time many of the corrective actions put forward to reduce compressed air energy consumption have been simplified with the intention of encouraging action. Although this is done with the best of intentions, sometimes simplifications and generalizations do not necessarily lead to positive results. It is the intention of this paper to highlight some of the more common issues that have been identified and not to discuss best practices that add value.

One of the most common energy conservation measures (ECM) for compressed air involves reducing system
pressure. Some others involve installing system controls, leak repair and variable speed drive motors. To simplify the structure of this paper and manage the scope effectively, discussion will be limited to the network pressure reduction ECM.


The benefits of lowering system pressure can be attributed to two separate actions with one being reduced pressure at the compressor and the second being a reduction in pressure delivered to production equipment. Each has value and can be implemented quite easily but the savings associated must be calculated accurately before attempting to invoke an action that is normally not openly supported by production.


Reducing the pressure delivered to compressed air consuming equipment and processes will reduce the
volume of air consumed by the system. The energy savings associated with network pressure results from a
decrease in demand but will only be realized if the compressors reduce power as result of the demand change. The portion of system compressed air demand that is associated with operating at a higher than
necessary pressure is referred to as Artificial Demand. This is a relative value associated with the current system pressure and a target pressure. In the context of this paper, the terms demand, supply and volume are all referring to a volume with respect to time (flow). If volume is discussed as a static value (hydraulic volume) it will be stated as such. Supply is flow from the compressors and demand is the flow of air consumed by various production and process constituents.


To calculate the artificial demand, divide the density of the air at the target pressure by the density of the air at the current pressure and then multiply by the current demand. This will represent the demand at the reduced pressure with the difference between current and proposed demand being the artificial demand. Another way to calculate artificial demand is to divide the absolute pressures in place of density. Absolute pressure is the gauge pressure plus atmospheric pressure. However, if all calculations are being performed relative to standard conditions (scfm), then the atmospheric pressure at standard conditions must be used. Currently, the values for standard conditions (scfm) are 14.5 psia, 68°F, 0%RH. Therefore, a 100 psig (gauge pressure) value would equal 114.5 psia (absolute). Using different conditions in compressed air calculations is a common error. As an example, one cannot measure volume flow using a mass
flow meter calibrated in scfm yet perform storage calculations based on an atmospheric pressure of 13.9
psia. If standard conditions are measured, it also extremely important for any compressors added to the system to be specified in scfm based on site conditions.


When calculating artificial demand, accurately determining the current pressure and what percentage of
the demand is influenced by the change is critical. If a piece of production equipment is regulated at a pressure below the target pressure and reducing pressure in the system does not change the pressure on the consumer side of that regulator, compressed air consumption for that piece of production equipment will not change. Inversely, if a reduction in system pressure causes the pressure on the consumer side of the regulator to drop, the volume consumed by that application will be reduced. The reduction in flow for that application will be based on the change in pressures after the regulator, not on the system pressure. Because artificial demand is a relative value, a 1-psi reduction in pressure will have a larger impact on volume in a 40 psi application than a 100 psi application. Localized pressure is one of the reasons generalized calculations for artificial demand can be smaller or larger than the implemented results. Consequently, artificial demand calculations should take into consideration point of use (localized) pressure changes, not just the average system pressure change based on the largest measured demand.

If pressure changes occur at point of use applications, the artificial demand equation would need to be summed for all different conditions. More specifically, artificial demand would need to be calculated for each unique pressure application based on the localized pressure change and the localized volume. This can be very difficult to accomplish and measure, but is also the reason many people simply apply the general calculation for the entire demand and hope it works.

Another important issue is to accurately determine the current or initial pressure. A common error is to use the highest observed pressure at the compressor discharge as opposed to the average system pressure in the network. At the discharge of the compressor, the air pressure is the highest because there are no friction losses associated with moving the air through filters, dryers and pipe. The appropriate pressure used in the calculation is based on where the air is being consumed. Discharge of the compressor represents where the air is being supplied. Also, if average demand is utilized for the volume, then average pressure should be used for the calculation, not the highest pressure. Typically demand changes with pressure for compressors with controls that utilize proportional logic where volume output is adjusted as a function of pressure. Assuming a constant supply volume and a fixed number of compressed air consumers operating, if demand does not change as a function of system pressure, there is no artificial demand. Figure 1 is an example of how demand changes as a function of system pressure. For this specific example, one compressor is operating using a load/no-load control where the compressor loads to full capacity at a lower pressure set point and then unloads at an upper set point at which time the compressor output goes to zero. When this data was recorded, plant demand was at a steady state. For the graph in FIGURE 1, pressure is on the vertical axis and time is on the horizontal axis. When the compressor is loaded, 100% of the compressor output is going into the system. A percentage of that air is being consumed by compressed air users, and the surplus air is held in the system as inventory. This is analogous to a stack of boxes kept in inventory. As you put more boxes onto the pile as inventory, the stack of boxes gets higher. As you start to use up inventory, the stack becomes smaller as boxes are taken away. Pressure changes in a compressed air system in a similar fashion, where pressure rises as excess air is stored in the system, and then decreases as air is removed. Whenever there is a difference between the rate of air entering the system and the rate of air leaving, the pressure will change. When the compressor reaches the unload set point (pressure), the compressor unloads and output goes to zero. At this time demand is greater than the supply and pressure falls as compressed air is removed from inventory. Looking at the graph, one can see the pressure rising and falling over time as the compressor loads and unloads. The graph in FIGURE 1 has a blue and red line moving together but separated by 0.65 psi. This is because pressures on the graph were recorded in two different locations, illustrating the difference between pressure where air enters the distribution network and the pressure at the furthest end of the facility. In this example the pressure loss across the system is only 0.65psi. Had the pressure difference across the network been significantly larger, pressure and load would need to be calculated for regions of the network since assuming demand is evenly distributed along the pressure gradient could create significant error in the artificial demand calculation.

The volume of surplus compressed air influences how quickly the pressure rises when the compressor is loaded. As the volume of surplus compressed air decreases, the rate of pressure rise in the network decreases. Notice that the shape of this graph is not linear. Had it been linear (straight lines, not curved) that would indicate the system has no artificial demand. The reason for the curve is associated with an increase in demand as the pressure rises. As the pressure increases, the total demand from the system increases. Since the amount of compressed air consumers has not changed, the difference is the artificial demand. To illustrate the difference, two lines are drawn tangent to the curve. Notice the change in slope on the curve as the pressure increases. For this system at this system pressure, all of the compressed air consumers were influenced by the same pressure. This can be confirmed mathematically because the difference in demand based on the two tangents to the curve equals the calculated artificial demand based on the total demand and the two pressure points. This in turn illustrates one easier way to determine artificial demand for a given system. Note that if the unload pressure set point for this compressor were raised another 10-psi, the compressor would run fully loaded without unloading because the sum of artificial demand and production needs would be equal to the total supply from the compressor.


An easier method of estimating artificial demand that does not sacrifice accuracy involves performing a test on the system. Provided the system has periods of reasonably stable demand, allow the pressure to decay down to the target pressure while recording pressure with data loggers. For this test, the volume of air supplied by the compressor must be constant and not changing based on some control input. If the output of the compressor must change during the test, you must be able to accurately determine the change in supplied volume so the artificial demand calculations can compensate for the change in supply volume. Perform the test several times and then review the data from the pressure data loggers. Flow can be calculated based on the capacitance (scf/psi) and pressure rate of change (psi/min).

To calculate pressure rate of change, draw a tangent line for each of the two points on the curve representing the current and targeted pressure values for the system. Pressure rate of change for each pressure point is the slope of the line in psi/min. This is best done by zooming in on the curve to improve resolution. Since the pressure decay can be steep, you may need to zoom in at each of the two desired pressure points. Slope for any tangent is then based on the pressure span (vertical) divided by the time span (horizontal) between the two end points of the tangent line.

Capacitance is a unit of measure based on the available system storage in cubic feet (ft3) divided by 14.5 psi if your calculations are in standard conditions. Since capacitance is being calculated based on standard conditions, it is commonly referenced as scf/psi. This is because standard flow is represented as scfm (scf/min) so standard volume (hydraulic volume, not flow) is referred to as scf, standard cubic feet. This should not be mistaken with standard flow which is measured in standard cubic feet per minute (scfm). Capacitance is the relationship between available compressed air from system storage relative to a 1-psi unit of change. This can be calculated by adding up the volume of all air receivers and piping in the system (in ft3) and dividing by 14.5 psi. There are several methods for calculating capacitance with one of
the common methods leveraging a calculation that relates pressure rate of change and supply volume to determine capacitance. The challenge with that method is an assumption that pressure rises and falls equally across all elements of storage. This is true for some systems however becomes complex when storage is installed ahead of dryers and filters. This wet receiver, as it is commonly called, will see a larger pressure change associated with a compressor loading and unloading because pressure drop across the dryer and filters will cause the upstream pressure to be higher than the network pressure. Another issue is caused by air receivers installed at the point of use. These smaller air receivers mounted
inside production equipment will typically be installed downstream of a check valve or regulator. Consequently, these air receivers can see a change in pressure that is different than the network pressure change used for the capacitance calculation. Since the air receivers frequently represent a significant percentage of the total volume, using this calculation method without accurately accounting for changes in pressure at each location can lead to significant error. For this reason, that calculation method has been excluded from this paper. Note well, the discussed method for artificial demand calculation only uses the pressure rate of change while the system is in draw down and demand has exceeded supply, at this point
pressure will drop at the same rate across the entire system with the exception of areas not influenced by the pressure change.

Considering the frequent use of flow meters, it must be stated that artificial demand cannot be measured with a mass flow meter when the system is in draw down. A mass flow meter estimates mass flow based on a calculation referencing heat capacity of the gas and energy relative to a temperature measurement at a point in the pipe. Neglecting any potential installation or velocity profile induced errors, mass flow for a fully loaded positive displacement compressor (rotary screw and reciprocating compressors) does not change as the discharge pressure changes. A difference in flow associated with air slippage and differential pressure
across the compression element may occur, but the difference is negligible. When the system demand exceeds supply (draw down) and pressure is falling, the supply from the compressors does not change even
though demand is changing. Consequently, a mass flow meter will read a constant supply and will not be able to account for changes due to artificial demand.


When the pressure is falling, demand is greater than the supply. Inversely, supply is greater than demand when the pressure is rising. A zero rate of change (no change in pressure with respect to time) occurs when supply and demand are equal. The difference between supply and demand is calculated by multiplying the capacitance (scf/psi) by the pressure rate of change (psi/min). The product of the two is expressed in scfm. This value plus the total volume flow rate from the compressors during the test equals the total demand. To calculate the measured artificial demand, subtract the calculated total flow at the lower pressure from the calculated total flow at the higher pressure. This can then be compared to the theoretical calculation based on total demand at the higher pressure value and the lower target pressure using the previously stated formula. If the measured artificial demand is smaller than the theoretical value, portions of demand are not influenced by the change in network pressure. To confirm, look for regulated consumers where an upstream change in pressure has a negligible impact on regulated pressure. If the measured value is larger than the
theoretical value, look for pressure change downstream of regulated consumers as the relative effect can be larger due to a lower initial pressure. For example, reducing pressure 4-psi from 100-psi will reduce the demand by 3.5%. Reducing the pressure 4-psi from 40-psi will reduce demand 7.3%. This is more than twice the impact because the initial pressure is only 40-psi, which is a common regulated value.

It is important to reaffirm that artificial demand is a reduction in demand, not energy. Assuming the pressure
at the discharge of the compressor does not change and network pressure was reduced using some type of
pressure reducing device, the energy reduction will be based on how the installed compressors reduce consumed power relative to the reduction in supply requirements. A best in class system will reduce power almost directly proportionate to the change in demand. Other systems will deliver a reduction in power that is a percentage of the demand reduction with the extreme being a system with centrifugal compressors that have no more throttle capability and are discharging excess air to the atmosphere in an effort to control pressure. For this type of system, a reduction in demand will have no impact on power. Fortunately, there are also situations where the power reduction is significantly larger than the percentage reduction in demand. Leveraging the centrifugal compressor example, if the reduction in demand is sufficient, a compressor can be turned off completely, eliminating air being discharged to atmosphere and increasing supply efficiency. This can also occur with rotary screw type systems.


There are several ways to reduce network pressure with the most common two being resetting the compressor pressure set points to a lower value and the second being installation of a system pressure reducing device. Common names for this type of device are demand expander, flow controller, pressure reducing valve (PRV), and regulator to name a few. Whether network pressure is reduced at the compressor or using an installed device, similar results can be achieved, each with its own advantages depending on the operating conditions. The impact of pressure change at the compressor discharge will be discussed in the next section.  An unfortunately overlooked strategy to reduce artificial demand is quite often the cheapest to implement. Focusing on tuning regulated users can deliver significant reductions in demand if a large volume of compressed air is consumed by regulated users. It is common to find regulators set excessively high for three major reasons.

    The operator thinks more pressure makes the application work better and has the ability to adjust the regulator without justifying their actions based on measurable process improvements.
    At some time, an issue occurred where compressor failure caused pressure to fall excessively across the network. When operators noticed the impact on process and the reduced pressure, instinctively they turn the knob on the regulator in hopes of increasing pressure even though it is a network wide issue. When system pressure is restored, regulators are not turned back.
    The application is installed with the regulator set at a specified value, not the lowest pressure required to deliver specified results. This is extremely common with air cylinders. Regulators will be set at 90-100 psi for a cylinder that will deliver sufficient force and speed at a significantly lower pressure. During installation and tuning of the installed equipment, the cylinder speed is reduced by increasing back pressure on the exhaust metering valve instead of lowering the pressure. The net result for example, is 95-psig supply with 55-psig of back pressure. Unless the 95 psig is required to deliver a specified force after the cylinder is fully extended, this same cylinder could be tuned to operate at close to 40 psig by adjusting the regulator and exhaust valve, reducing the required volume 49.3%. The higher regulated pressure may also be required to compensate for undersized components causing excessive pressure drop while the cylinder is extending or a poorly sized regulator. This can often be seen by watching the pressure gauge when the cylinder strokes. When friction (pressure drop) ahead of the regulator or the regulator itself is the issue, the gauge reading will initially drop while the cylinder is extending and then recover after the cylinder is fully extended and flow has stopped. At this point, the work has been done and air is flowing to the cylinder for no purpose other than to raise the pressure unnecessarily, increasing the consumed volume. When there is no flow there is no friction loss and pressure loss goes to zero. It is not uncommon to see 30-50 psi deflection that can be corrected to reduce demand. Although this action may not be as glamorous as a demand expander with a segmented valve and PID control, it can be implemented with almost no capital investment and can deliver significant results for many systems. There are a large number of facilities that have reduced total compressed air consumption 20- 40% by diligently tuning point of use applications. An important closing note on this topic: Record the measured values for future reference and place a laminated tag with the settings on the regulator. Installing pressure taps (Pete’s Plugs are a common example) at gauges and critical pressure points will facilitate quick validation of pressure settings and gauge accuracy using one calibrated mechanical or electronic pressure measurement device.


A pneumatic cylinder is an extremely common consumer of compressed air found in a production facility. They are incorporated to perform many tasks and the symptoms of failure can be summarized into two categories, insufficient speed of movement and insufficient applied force. FIGURE 2 illustrates a typical installation of a pneumatic cylinder. The force applied by the cylinder is a function of the pressure acting on the surface of the piston in the cylinder. As air enters the cylinder, the pressure rises until the applied force can overcome the forces acting against the cylinder and the internal friction losses. This pressure is referred to as the article pressure and is the critical pressure required in the cylinder to have movement. The rate of flow (scfm) will dictate the speed of the cylinder. Resistance to flow across the in-line components (filters, regulator, tubing, etc) will limit the rate of flow and therefore limit the rate (speed) of the piston moving. Consequently, if the performance symptom is a reduced cylinder speed, the potential problem could be the result of an increase in resistance across the pneumatic supply components. Aside to changes to components or mechanical damage causing increased friction, an increase in resistance is normally associated with dirt loading in a filter.


If the symptom is insufficient force, the pressure acting on the piston in the cylinder (article pressure) needs to be considered. Pressure drop across a resistance changes as a square function of the flow. When the flow goes to zero, there is no pressure drop and the force exerted by the cylinder is not affected by the inline components. However, if there is a leak after the components, there will always be flow and therefore a pressure drop across the components. This will limit the applied force from the cylinder. As the size of the leak increases, the loss in pressure increases until the pressure loss across the components reduces the cylinder force enough to negatively impact the application. The leak could be in fittings, worn tubing, failing piston seals, or other areas. A 1 scfm leak on the above mentioned 1.5” cylinder could easily quadruple the pressure loss to the cylinder. As the leak develops, the operator will compensate by increasing the pressure from the regulator. When the regulator has been adjusted to the highest possible setting, they will usually request that system pressure be increased in the compressor room. Eventually, the leak will be noticed and repaired. Without the continuous flow across the point of use components from the leak, pressure losses to the cylinder will return to their initial values. Although the required pressure to operate the cylinder can now be reduced, they are typically left at the elevated value, redefining the pressure requirements for the system. Situations like this cause compressed air systems to operate at pressures much higher than required, increasing artificial demand and wasting energy.


As previously stated, lowering pressure in the network can potentially reduce artificial demand, impacting the demand for compressed air. Needing less compressed air typically translates into reduced compressor power. However, when reducing pressure at the discharge of a compressor, a secondary benefit to the reduction in artificial demand is the potential impact on compressor power and sometimes capacity. To describe the benefits of lower pressure at the discharge of a compressor, it is best to differentiate between positive displacement and dynamic compressors.


The best way to describe a positive displacement compressor is by thinking about a good old fashioned bicycle pump. At first when pressure is low, moving the handle up and down is easy. As pressure increases, it becomes more difficult to push on the handle. If the pump has a 2 inch diameter piston, one would need to put all of their weight on the handle to push 60-psi of pressure into the tire. This is because work is done against the pressure at the discharge of the pump. With a 2 inch piston and a surface area of 3.14 in2, the force acting against the piston is in excess of 188 lbf. A positive displacement compressor is very similar because for the most part, discharge pressure is based on the pressure in the system. Considering this fact, the work required to move a given volume of air through the compressor is reduced as the discharge pressure is lowered. For a positive displacement compressor like a rotary screw compressor or a reciprocating (piston) compressor, this change in pressure will result in approximately ½% drop in power for every 1-psi reduction at the discharge. Although this rule of thumb is reasonably close for most positive displacement compressors, it does have limitations. Ideally, the manufacturer should always be consulted to determine their anticipated rerated power. Unfortunately this information is not always readily available or a sales person with the best of intentions simply multiplies ½% by your stated pressure reduction to give you a new number. Considering the effort required to gain support for pressure reduction projects, it is very unfortunate when projects move forward and anticipated savings are not realized. To help gauge if predicted savings are realistic, the influence of the pressure reduction on the compressor must be considered.


It is common to see calculations that assume ½% less power for every 1-psi reduction in pressure at the discharge of the compressor, regardless of initial design pressure. If this theory held true, manufacturers would design compressors for 300 psig and then run them at 100 psig with 0% power. This is obviously not the case and internal resistance is one of the reasons. For calculation purposes, a rotary screw compressor designed for 100 psig discharge pressure at full flow will be used. This is one of the more common types of compressors used for industrial applications. Consider the difference between pressure at the discharge of the compressor element and the compressor package. For a contact cooled (wet or oilflooded) screw compressor, air must go through internal piping, an air/oil separator element, baffles, minimumpressure check valve, heat exchanger, and a moisture separator before exiting the package. In an effort to keep things simple, let’s assume 15 psi of pressure loss across all these components with an average pressure loss across the air/oil separator element. If pressure at the discharge is reduced to 20 psi, the anticipated reduction in power would be 10%. However, as the air is expanded to a lower pressure, the density of the gas is reduced. Assuming constant mass flow at full load, the volume of air moving through the compressor at 80 psig is greater than the initial volume at 100 psig. Calculating the difference based on the absolute pressures, volume should increase approximately 1.212 times. Since cross sectional area does not change for the compressor components, the velocity would also increase by a factor of 1.212. Since pressure drop increases as a square function of the change in velocity, the pressure drop would increase 1.47 times and the 15 psi pressure drop across the compressor package components would now be closer to 22 psi. Considering the 20 psi reduction at the discharge of the compressor and the 7 psi increase in pressure loss across internal components, the compressor element would only see a 13 psi reduction in pressure, not the 20 psi that would be seen at the discharge of the compressor. Consequently, the power reduction would only be 6.5% as opposed to the assumed 10% if only the (1:2) rule of thumb calculation were considered. The velocity influence on heat exchanger performance would compound velocity and pressure drop issues, deflating the savings projections even further. There are also velocity constraints that may be an issue in some compressors since air does not have an unlimited velocity.


There are two components in a rotary screw compressor that limit how power will change when pressure at the discharge is reduced. The minimum pressure check valve prevents the sump from filling to line pressure when the compressor is unloaded or off. The valve also serves to maintain a minimum sump pressure to prevent excessive oil carryover from high air velocity across the air/oil separator. This minimum pressure check valve limits pressure reductions to the compressor element when pressure is lowered at the compressor package discharge.

A rotary screw compressor has a minimum discharge pressure in the design of the rotors, stator and discharge plenum. This is based on a compression ratio in the design and how air exits the rotors through the discharge plenum. Based on design, the air end (compressor element) will have a minimum pressure generated internally. As pressure deviates from design pressure, the net effect is dampened by the compressors need to build a minimum internal pressure based on the design. Consequently, the influence on power diminishes as the discharge pressure gets further away from internal design pressure.


The most accurate method of confirming the change in power associated with a reduction in pressure is to measure it. Run the compressor during a night or weekend shift when demand is lower and system pressure can be reduced. If this is not possible, drop system pressure carefully during normal production and note the change in power. Ideally, actual power (kW) should be measured, not apparent power (kVA). Apparent power is normally calculated by measuring amperage and voltage. Utility companies charge based on actual power in kW so appropriate testing equipment is essential. The need for kW meters is due to the change in power factor associated with a reduction in motor load. The power factor is also influenced by external factors in the system. This will cause the system power factor to change based on other loads placed on the system. As the power factor value becomes lower, the amperage will increase, which misrepresents the actual power. Power factor can change independent of the compressor operation, thus misrepresenting a change in compressor power.


It is extremely common to see energy assessment specialists treat centrifugal compressors like positive displacement compressors when it comes to energy reductions associated with pressure change. Unfortunately, this is not the case and estimated energy savings are not being realized. Since centrifugal compressor systems are normally larger, the miscalculations represent hundreds of thousands of dollars. These errors are not done maliciously; they are unfortunately the result of oversimplified best practices perpetuated by individuals with limited centrifugal compressor knowledge. This type of knowledge is not readily available as most energy assessment specialists do not have access to engineering teams responsible for the technical development and design of centrifugal compressors. From a unit perspective, centrifugal compressors are a small part of the compressor market so technically knowledgeable resources are limited.

Unlike a positive displacement compressor, centrifugal compressors cannot increase their pressure capabilities by increasing power. As introduced in the previous example regarding the bicycle pump, work and power were related to forces acting against the volume of air being discharged to the system. This is because air is pressurized by compression or forcing a given mass of air to occupy a smaller space. A centrifugal compressor generates pressure in a different way. A given mass of air is accelerated through an impeller, imparting kinetic energy. The air goes through a diffuser where the air is slowed down, converting kinetic energy in potential energy. This manifests itself in the form of increased pressure and heat. Depending on pressure requirements of the compressor, air will go through the same process through subsequent stages, building towards the design pressure requirements. To improve efficiency, some or all stages will cool the air before entering the next stage.

For a given compressor design, the maximum pressure is limited by surge. At surge, a flow reversal will occur and compressor operation becomes unstable. Consequently normal operation attempts to limit surge through the use of controls and measured variables such as amperage or Polytropic head. Polytropic head is the energy in foot pounds required to compress (Polytropically) one pound of a given gas. The intention of this paper is not to discuss the merits of surge prediction based on motor amperage or the calculation of Polytropic head. For discussion purposes and an effort to remain within scope, pressure limitations for a centrifugal compressor are limited by design and thermodynamics, not power.

The relationship between flow, pressure and power for a centrifugal compressor are normally expressed using a performance curve. The pressure and flow capabilities of the compressor are influenced by ambient conditions, cooling water, and mechanical condition of the compressor. As a result, the performance, and most notably pressure capabilities will change as ambient conditions change throughout the year. To help better illustrate this effect, a working curve is utilized that has curves for three sets of ambient conditions on one plot. An example is referenced as FIGURE 3.


The performance curve is made up of two parts. The pressure – flow curve has pressure on the vertical axis and flow on the horizontal axis. The power – flow curve has power on the vertical axis and flow on the horizontal axis. The flow values for each horizontal axis align so each pressure – flow curve has a matching power – flow curve. Notice how the natural curve moves up and to the right as ambient temperature decreases. Looking at the red curves for power and pressure with respect to flow, moving from left to right, a vertical line intersecting both curves would illustrate the design pressure and power for that specific flow and ambient conditions. Moving from left to right, notice how the power initially increases as pressure goes down and then decreases as one moves further to the right. This illustrates how power is not directly proportionate to a change in pressure. Once again referencing the red pressure – flow curve, as pressure decreases flow increases for the compressor. Flow will increase as pressure is reduced but notice how the slope of the curve changes as pressure decays. Eventually the curve will become asymptotic (straight up and down) when the compressor moves into a region known as choke or stonewall. At this point, dropping pressure will have very little to no change on flow or power. This describes the first major issue with applying the ½% power per 1 psi rule to a centrifugal compressor. At pressures at or below choke, power will not decrease.

Looking at performance within the active part of the curve, changes in pressure have an impact on flow and power. To better illustrate the impact, data for the curve is listed in FIGURE 4.


The data in FIGURE 4 is based on tested performance for a specific centrifugal compressor. Looking at compressor performance at 121 psig and 111 psig, reducing pressure from 121 to 111 psig would only lower power 5 horsepower. This represents less than 0.35% reduction in power. The ½% per psig rule obviously does not apply as it would have predicted a 5% reduction in power with an estimated savings of $50k per year as opposed to the $3k that would have been realized. However, since the compressor in this example is operating within the active range of the curve, flow increases ~100 scfm in this example. Assuming demand stays the same and the compressor power changes directly proportionate to the change in flow, the compressor would have reduced power by 27 horsepower or 1.8%. This is still less than 36% of the savings that would have been estimated using the ½% per psig rule of thumb.

It is important to note that unlike rotary screw compressors, centrifugal compressor model numbers do not necessarily represent performance of the compressor. For a given external casting, design and motor; several different impeller/diffuser combinations may be used. The combination of impellers and diffusers is commonly referred to as the “aero” for the compressor. For a given compressor model number, several different aero packages can be used, each with its own unique performance. One CANNOT use a generic curve or even a curve from the same model of compressor unless the manufacturer has confirmed that the compressors were manufactured using the same aero. It is equally important to ensure data is corrected for site conditions or a range of conditions if ambient changes with respect to time. Referencing FIGURE 3, the three curves from left to right represent data from 95°F, 70°F, and 30°F conditions. Based on how the performance curve shifts relative to temperature, it is not uncommon to find compressors that operate in choke for several months a year. This is significant since any estimates for energy savings associated with pressure must take into consideration time, temperature and location on the curve. Without this data, any attempt to estimate savings associated with pressure may be misleading. 


Maintenance and more specifically the mechanical condition of the compressor can also influence the performance and the shape of the performance curve. Depending on atmospheric conditions, over time the compressor aero can become dirt loaded, pitted, or eroded. Degradation of the aero may impact performance significantly before impacting vibration levels. Even if the compressor is routinely overhauled or serviced based on some other fault, impellers are not necessarily replaced or machined back to original specifications. It is not uncommon to find low cost service providers that will simply grind and balance an existing component or replace it with one of a limited number of options they have available. Compressor vibration will be within specification but performance may change substantially. To confirm compressor performance, the compressor should be surge tested routinely. This is usually done at least once per year. Surge testing would consist of testing for natural surge (top of curve), throttle surge (far left point at pressure) and maximum power at pressure. This defines the 3 points on the curve an must be compared to the engineering data from the manufacturer. To compare effectively, engineering data for the specific aero must be corrected to the site conditions during the surge test. Provided the compressor is fine, measured data should be close to engineering data. If there is a discrepancy but vibration is not an issue, continue to review tested data against the engineering data to identify a trend. With this information, compressor service can be scheduled at a convenient time before mechanical integrity of the compressor becomes an issue or the compressor is not capable of supporting production requirements. This is also a good way to review performance after an overhaul to ensure performance has returned to back to initial design. Surge testing the compressors is also a good way to confirm the performance curve data for a given compressor is correct. If not, any analysis done to estimate energy savings associated with pressure reduction may be incorrect.


Maximum pressure capabilities of a specific compressor are based on the aero package, ambient conditions, and mechanical condition. The maximum operating pressure is limited by the compressor surging at the top of the curve. This point is called the natural surge pressure. Referencing FIGURE 3, the pink horizontal line
represents the constant pressure line. When demand is less than the maximum flow from the compressor, the inlet will throttle to reduce flow. With inlet guide vanes, the efficiency will remain reasonably constant while the compressor is throttling. The throttled power can be seen on the lower power – flow curve as the diagonal line. The compressor can reduce the output flow a limited amount. Following the pink horizontal line in figure 3 to the left, the minimum stable flow is dictated by the point where the constant pressure line intersects the throttle surge line. If the compressor attempts to limit flow to less than this intersection point, the compressor will surge. For obvious reasons, this is called a throttle surge. The throttle surge line can be seen on FIGURE 3 as the blue diagonal line on the pressure – flow plot. If the demand for air is less than this minimum constraint, excess air is discharged to the atmosphere to compensate for the difference between minimum stable flow and demand requirements. Unfortunately, after the compressor stops throttling, the
power does not change. Consequently all the air that is being discharged to atmosphere is being wasted. For a compressor that is operating frequently with air bypassed to the atmosphere, lowering pressure will reduce the flow where throttle surge will occur. As a result, after adjusting control settings a compressor operating in choke could still reduce power by increasing throttle capabilities by dropping pressure. This is only the case if the compressor is bypassing air to the atmosphere and the controls can allow the compressor to close the inlet, increasing throttle capabilities and decreasing power. Once again, the site corrected performance curves are required to quantify the potential savings. Also, the ability to operate a compressor close to throttle surge is limited by the complexity of the control algorithms, throttle variable used and how the compressor PID loops are tuned relative to system dynamics. FIGURE 3 illustrates the power reduction associated with adjusting the compressor throttle limit from a conservative setting to a more efficient setting by simply tuning the compressor PID loop so compressor reaction rates matched the rate of demand changes effectively. From figure 3, looking at the two vertical dashed dark red lines from the constant pressure line to the throttled power line, the change in power associated with adjusting controls can be seen. For this compressor, power was reduced 160 hp with no capital investment. The compressor still operated bypassing air to atmosphere, but the amount was reduced from an average of 1,200 scfm to 220 scfm.


For every compressed air system, regardless of types and brand of compressors, the opportunities available to reduce costs and improve quality exist. However, the method of achieving the end results varies from plant to plant. For some facilities, changes to the control philosophy and equipment in the compressor room will represent an opportunity for savings. For others, all of the opportunities for improvement are based on how compressed air is used within the production area. The common issue that exists in almost every facility is associated with a lack of system metrics and visibility to the cost of operation. Rarely do systems instrument performance variables such as actual power, volume and pressure. These variables need to be monitored and trended relative to production metrics.


Simply having performance metrics for the compressed air system is not sufficient; someone needs to be held accountable for the performance. More specifically, without visibility there is no accountability. An individual or group must be responsible and accountable for any increase in cost associated with operation of the compressed air system. It is important that the performance metric be based on total kWh and total
volume of compressed air with respect to production output. If production is measured based on number of
units per shift, compressed air performance should be measured as total power and total volume relative to the same production units. Measuring the efficiency of compressed air supply (scfm/kW or kW/100scfm) will not account for waste in the system. As an example, a best in class system will maintain peak efficiency at any demand. Consequently, if waste increases in the system (leaks as an example), supply efficiency will still be great even though an increasing percentage of the air being produced is discharged to atmosphere with no increase in productivity. Ideally, this performance metric should be owned by production and facility management. Otherwise, the consumers of compressed air will not share in the responsibility of managing consumption and waste. Without these metrics and ownership, any efforts to initiate changes in operating philosophy or consumption will fail or not be sustained. Any project to improve the operating efficiency of the compressed air system must start with system instrumentation and metrics.

It is common for compressed air optimization efforts to take place as events with a focus on new equipment and capital investment. This would also include identification of leaks and a reduction in supply pressure. Since corrective actions are identified and implemented as a project (an event), the results are not monitored or tracked over time. Consequently, at the first complaint regarding pressure on the production floor, system pressure is typically raised along with the associated compressor operating cost. Leak load increases without measurement and the cost for compressed air increases. The number of running compressors and their operating protocol is rarely defined or reviewed. Without guidelines, any adjustment to compressor settings is acceptable, potentially decreasing supply efficiency. All of these issues could be identified and addressed provided instrumentation and metrics were leveraged to maintain desired results.


Reducing system pressure remains an effective energy conservation option, provided sufficient effort, experience, and knowledge are involved in estimating the associated cost savings. Simple assumptions and rules of thumb do not apply and can lead to poor decisions with negligible energy savings. With some compressors, reducing pressure may have no impact on compressor power. In these scenarios there are no cost savings at all.

Accurate analysis and detailed planning of corrective actions is essential to delivering predicted results. However, without appropriate performance metrics and ownership, energy savings will degrade over time.