Driving change in chiller design

RELATED ARTICLES: Latest chilled-water system from Trane | Trane keeps cool in heated market place | Trane Man

A seasoned veteran of the HVAC industry, Don Eppelheimer has been working for Trane since 1972. As global chiller systems manager: commercial systems, he recently hosted a workshop on chilled water system design in Dubai, sharing his years of expertise with an appreciative audience.

Eppelheimer began with a brief overview of the main components of a centrifugal chiller: “There is an evaporator where the evaporation of refrigerant cools water. In the back is the condenser where hot refrigerant vapor is condensed so we can reject the heat to the cooling tower. On the top you see a two-stage centrifugal compressor and motor to spin the impellers. In the US, and in Saudi, the impellers spin at 3 600 rpm. Here in Dubai and in the UAE, the impellers spin at 3000 rpm. We need a starter to control the motor, and we need a control panel to provide the operator with the necessary level of access.”

As to how the centrifugal compressor itself works: “There are two impellers mounted on the motor shaft. It is, of course, the spinning of the impellers that compresses the refrigerant. Upstream of the first impeller, there are wedge-shaped inlet-vanes. A centrifugal compressor works by spinning the gas, which creates velocity. The gas enters through the eye of the impeller and the spinning of the impeller sends it out through the circumference. On leaving the impeller the refrigerant enters the diffuser, where it decelerates. So gas accelerates in the impeller, and decelerates in the diffuser.”

Eppelheimer pointed out that this progression can be plotted. “When the refrigerant velocity decreases, velocity pressure is converted to static pressure. It is a combination of the impeller, the diffuser and the rotating action that allows the centrifugal compressor to create pressure.”

“A couple of features to remember about the centrifugal compressor: it is the inlet vanes that control the volume, which becomes (tons of) capacity. So it is inlet vanes that control part load. But it is the tip speed – the velocity of the impeller – that creates lift.” However, it is mass flow of the refrigerant through the compressor, in pounds per hour that becomes tons. “The pressure created by a centrifugal fan is static pressure; the pressure created by a centrifugal pump is head, and the pressure created by a centrifugal compressor is lift.”

“Lift has units of temperature. Because we are compressing a chemical with the unique property of saturation, for every pressure, there is a corresponding temperature. This makes it easy for HVAC engineers to use temperature to discuss pressure, as the type of refrigerant is not an issue. It can be a low-pressure refrigerant where the units are very low; it can be a medium or, even a high-pressure refrigerant, but the application is still the same when we discuss temperature.”

In respect of the work that must be done by the compressor we must consider leaving chilled-water temperature, and the temperature of the water leaving the condenser. Leaving condenser water temperature, minus leaving evaporator water temperature, is what we call ‘lift’. We adjust the load with inlet vanes, and we adjust the lift with tip speed. When we talk about applying drives to centrifugal chillers, a unit of measurement called NPLV, APLV or IPLV often surfaces. .”

Eppelheimer explained that IPLV is a mathematical definition of part load, created by ARI (Air-conditioning Research Institute) for the benefit of ASHRAE. ASHRAE Standard 90 assumes minimum levels of performance for various pieces of equipment in air-conditioning systems. When it came to chillers, the authors of ASHRAE Standard 90 highlighted the importance of full load. “Full load determines electrical line sizes, electrical distribution losses, the size of transformers and their efficiencies, and the impact on society: i.e how many chillers, how many power plants need to be built and what their efficiencies are going to be; thus full load is very important.”

Equally important, however, is part load. “Part load efficiencies determine carbon footprint or, how much energy we consume over a year. Therefore, ASHRAE Standard 90 imposes minimum levels of performance for chillers e.g. at over 300 TR, the flow load performance has to be a COP of at least 6.1 at ARI conditions. The part load performance, or IPLV, needs to be at least 6.4.” This required an easily understandable and implementable definition of part load, which ARI was tasked to devise.

Eppelheimer said ARI looked at four different load points. For example, the first load point was 100%, with an entering condenser water temp of 85˚F. “We compute the energy that the chiller consumes at this condition, and it becomes 17% of the value. The second load point was 75%. The assumption was that, at 75% load, the entering condenser water temperature was 78.75˚F.” What was important was that all four of these load points “had different weighting factors as to their impact on APLV.”

In 1998, the ARI Chiller Committee created a new standard and measurement. “The chiller manufacturers voted to improve the standard, so APLV was replaced with NPLV.” Eppelheimer said differences were immediately apparent. “The weighting factors at 75% and 50% load were increased; the weighting factor at full load was almost eliminated. Basically, they took the term ‘load’ and tried to make it narrower.” This begs the question as to why APLV, which had proved to be a very useful tool, was ultimately replaced with NPLV.

“Let us say we have a 1 200 TR chiller. We can equip that chiller with a standard starter, or with a drive. As we are looking at the following load points: 100%, 75%, 50% and 25%, what is really important is what the assumed entering condenser water temperature is at those points.”

“Note that when the entering condenser water temperature is 85˚F, the drive, due to electrical losses, consumes about 30 kW. When the cooling tower water temperature falls to 75˚F, , the drive saves about 30 kW. So, between 85˚F and 75˚F, the benefit of the drive is negligible. To achieve substantial savings, by putting an inverter on a centrifugal compressor, we need to reduce the lift. When the condenser water temperature is 65˚F, you can see some substantial savings (over 70 kW at 50% load), while at 30% load; the drive reduces chiller amps by a third.”

Does APLV take such findings into account? “APLV states that, for almost 20% of the year, the drive is losing energy. Yet at 50% load and 65˚F, where there is substantial savings, it counts for only a third of the APLV value. Can those weightings be changed? Of course, APLV is an arbitrary definition of chiller part load. So in 1998 ARI replaced APLV with NPLV. ARI took the full load percentage point, where drives are not very attractive, and reduced the weighting to 1%. They took the 50% load point at 65˚F, where drives save a substantial amount of energy, and increased the weighting to 45%. By adjusting the weighting factors 100% load was practically erased.”

However, Eppelheimer said this had an unintended consequence. “Chiller manufacturers, owners, and operators have a fixation on tons. If you were to go into a chiller plant and ask the operator, how often do you see your centrifugal chiller running at 100% load? Well, it never happens. How often do you run your chillers at 25% load? Well, they are not very efficient then, so we try and avoid it if possible. We spend most of our operating hours between 50% and 75% load.” The weightings of NPLV conditions seem credible when viewed as load.”

The unfortunate flipside of this was NPLV’s related assumption about cooling tower performance. “NPLV says that the tower is at 30˚F for 1% of the year, 24˚F for 40%, and 18˚F for almost 60% of the year.” So, does this hold true for Dubai? Eppelheimer said Trane used its Trace700 chilled water modeling program to determine “how much energy the chillers, pumps and fans might consume. Note that that the program takes all the components of our chilled water system and runs that building in a computer model for every hour of the year.”

The outcome of this simulation was that “here in Dubai we actually have a population of hours where the entering condenser water temperature is warmer than 85˚F, and even up to 95˚F. This is hot.” There are obviously also occasions where the load is reduced, meaning that cooling tower temperature could be reduced as well. “This is 6% of the time, according to our Trace700 simulation. For 6% of the time, the chiller load is 25% or less, and the cooling tower water is 65˚F or colder. IPLV states it is 12% of the time. Yes, this is a good thing, but IPLV has overestimated a good thing by a factor of two!

“What about 65˚F cooling tower water but 50% load? This is where drives really save energy. According to IPLV/NPLV, drives operate here almost half the time, at 45%; unless you are in Dubai, where it is only 2%. So for this condition, NPLV is overstated by 22.5 times.” What is the solution to this problem? Eppelheimer said that while the benefit of applying drives to fans and pumps was well known, “how do we really determine the benefit of drives on chillers?”

He outlined three alternatives to illustrate this:
• Drives on a cooling tower, with a cold setpoint;
• The same system, but with a control change: instead of controlling the cooling tower for a cold temperature, we are going to control the cooling tower for a warmer temperature. This will increase the savings on the cooling tower; and
• A cold cooling tower, but taking advantage of variable speed drives on the chiller.

Eppelheimer said the immediate result from such comparative simulations was that “the variable speed drive centrifugal chiller always saves energy. We would expect it to be the most efficient chiller. But in order to take advantage of that efficiency of variable speed, we need to reduce the lift. And to reduce the lift, I need more cooling tower fan energy.”

“So when you look at the potential savings on the compressor by using variable speed, you also have to pay attention to the potential savings achievable on the cooling tower. It is not really possible to do both, however, a controller cannot follow two masters. This means a single strategy; and that strategy has to be focused on realizing savings on the chilled water system, even though the compressor is the largest motor.”

Looking more closely at the simulation, Eppelheimer said that, from January to March, “the variable speed chiller system consumed the least amount of energy. For the remainder of the year, the variable speed cooling tower saved more energy than the variable speed chiller. We might expect that because Dubai is a very hot climate. But this is not unique. We have found similar results even in the US, which is much cooler. It seems to be a recurring trend that a variable speed cooling tower saves more energy in the summer months, while the variable speed chiller saves more energy in the winter months.”

It is important to note that NPLV is used to show that the chiller you are considering purchasing complies with ASHRAE Standard 90. In order to be ASHRAE Standard 90 compliant, it has to exceed minimum full load efficiency, and it has to exceed minimum part load efficiency. That was the intent of APLV/NPLV. However, we do not think it is appropriate to use it to evaluate the performance of an entire chilled water plant.”

This is because the definition was compiled initially “for a single chiller, and now the industry is using it to make decisions or evaluations of a chiller plant that might have multiple chillers.” Another important point to be considered is the impact of climate; “Which is an excellent argument, in my opinion, for not using NPLV, as this regards the climate in Dubai as being exactly the same as in Chicago, for example. I do not think it is the number of chillers that is important; it is climate that is important.”

Don continued, “It is the leaving water temperature, not entering that affects compressor load. So, in the definition of NPLV, it is presumed that the condenser water flow is 3 gpm/t. Why does ARI rate chillers at 3 gpm/t and 44˚F? Well, as a chiller manufacturer we can claim its capacity at 44˚F is 1 100 TR. However, its capacity at 40˚F deg is only 950 TR. Now would a chiller manufacturer want to tell you this is a 950 TR chiller or an 1 100 TR chiller?

“At a condenser water flow of 3 gpm/t, the COP is 7.1. At a condenser flow of 2 gpm/t, the COP is 6.8. Again, would a chiller manufacturer, prefer to list its chiller as a COP of 6.8 or 7.1? Thus in terms of the standard ARI conditions, maybe we should strike the word ‘standard’ and replace it with ‘ideal’, as it overstates the applied capacity and COP of the chiller, based on warmer than applied water temperatures generally, and greater than applied condenser volumes.

“If the entering condenser water temperature is 85˚F, and the flow is 3 gpm/t, the leaving condenser water temperature is about 94˚F. At 3 gpm/t, we get a 9˚F deg condenser Delta T. At 2 gpm/t, the Delta T on the condenser is about 14˚F. So at 2 gpm/t, the leaving condenser water temp is 5˚F warmer – of course, the lift is greater and the COP is going to be reduced. ASHRAE Standard 90 actually assists with this scenario. If you select a chiller that is colder than ARI ideal conditions, or condenser flow that is less than ARI ideal conditions, the permitted COP you have to meet to comply with in terms of ASHRAE Standard 90 is actually reduced.”

This is because the fundamental understanding of what is ultimately required is not necessarily the most efficient chiller. “We want the most efficient building. If you can save pump energy by pumping less water through the condenser, if you can save pump energy by pumping colder water out of the chiller, ASHRAE Standard 90 will give you credit for that. So the minimum COP full load and part load that you have to meet, at applied conditions, is lower.”