June 2008

Special Section: Temperature/Pressure

Ontario nuclear plant has a chilling solution

The problem is not enough differential pressure between the condenser and evaporator to allow refrigerant/oil circulation

Fast Forward

  • Management is not keen on bucking a manufacturer's recommendations.
  • It is difficult to get an off-the-shelf calculation for a complex system.
  • Hand calculations followed by testing gets us closer to the physics of the system.
By Ken Dias

Our utility has embarked on a chiller replacement program to meet environmental regulations.

Three service wing chillers come under this replacement program. The new chillers have dual screw compressors that are less forgiving in cold temperatures than reciprocating or centrifugal chillers.

The cooling water, which comes from Lake Ontario, can be as cold as 33°F (0°C) in winter.

The problem here is there is not enough differential pressure between the condenser and evaporator to allow refrigerant/oil circulation to prevent damage to the chiller especially during starting.

The manufacturer stipulates the new chiller reach a minimum of 55°F (13°C) within 15 minutes of starting. Hence, there was concern the chillers would not be able to operate satisfactorily under these process conditions.

Know these chillers are unlike most commercial chillers in that they do not use cooling towers. Lake water, and in effect the vast reservoir of Lake Ontario, are the cooling tower-the heat transfer element instead.

Existing chillers are out

Initially, the valve TCV1 is closed, which allows the water to re-circulate through the tank, pump, and condenser, picking up heat from the condenser and pump.

This heating of the cooling water under cold starting conditions is important for optimal refrigerant/oil circulation.
However, centrifugal compressors are more tolerant of low temperature operation, and the original system worked fine for close to 30 years. A pneumatic control system in a cascade loop controlled refrigerant head pressure by modulating cooling water flow.

A pressure controller worked as a PI controller to control the head pressure, its output working as the set point for a temperature controller used as a P controller, which controlled flow through the control valve.

The manufacturer came to our site to give a presentation on the problem and their proposed solution. Their proposed scheme is here. It entailed adding a flow diversion loop on the cooling water system.

They recommended the use of a 3-way valve to divert the condenser water flow, thereby increasing the rate of temperature rise. There are two separate refrigeration circuits and one condenser for each circuit.

Hence, there are two condensers, and these are set up in series in our chillers. They transfer the heat from the refrigerant to the cooling water, thereby heating the cooling water.

At low temperatures, the cooling load is low, leading to low cooling water rates in the condenser. Therefore, during start up and under cold temperature conditions, the 3-way valve would be partially open, diverting some of the water away from the condensers.

This means only a portion of the water goes through the condensers, allowing for more rapid heating. Once the temperature has reached the stipulated value, the 3-way valve is fully closed and normal cooling water flow resumes.
The minimum cooling load for stable operation of the chiller is 15% only. At low temperatures, as during the severe Canadian winters, the cooling load is at the minimum value.

Hence, a "double whammy" occurs when the lake water temperature is at its lowest and the need for heating condenser-cooling water quickly is at its highest. To add to the control complexity, the condensers are in series thereby pre-empting independent temperature control of each condenser.

Doubts, considerations

The I&C design people expressed concern on controllability due to the flow diversion loop and the associated additional components and decided to review the process and control system in detail.

This phase consisted of examining the existing system closely and comparing it with the vendor-proposed system, especially during cold-start conditions.

This was more of a qualitative comparison, as there are advantages and disadvantages in both schemes. The existing scheme has the advantage of being less complex, but there was concern the full effect of 165 GPM of cold lake water gushing through the condensers would cause problems.

The new scheme would prevent this by bypassing water away from the condensers through the 3-way valve. However, any failure of the 3-way valve or associated controls would damage the condenser, as they would not get any cooling water.

In addition, low cooling water flow rates increase the fouling rate in the condenser although some could suggest that this low flow rate takes place for only a short period.


There was an attempt to simulate the system using popular simulation packages that are available, but this proved futile, as an integrated package that could simulate the system was difficult to find.

We considered the following factors in our decision-making process:

  1. The chillers and associated cooling water piping are located indoors. Since this problem occurs during starting of a chiller, one could assume the stagnant water in the chiller that starts the process would have attained the ambient temperature of the indoor location and would then be at a higher temperature (say by 10°F) than raw lake water temperature.
  2. The manufacturer stipulates that a minimum pressure differential between compressor suction and discharge is necessary to move oil for cooling the compressor rotor and lubricating the bearings. Oil pressure differential is also required to load the compressor. Low temperature and pressure in the condenser and high temperature and pressure in the evaporator can reduce the system differential pressure below the minimum required for proper cooling, lubrication, and loading. The requirement is a 35-psid system differential pressure must happen within 15 minutes of startup. This corresponds to the 55°F temperature mentioned earlier. How does one ensure this transpires and prove it is in fact the case? Calculations involving refrigerants and water are complex-especially so, as the refrigerant flow is a 2-phase flow. These calculations would have to happen for the worst-case scenario, which is to say for initial starting under minimum load of 15%.


Approach and suitability

As it turns out, the time for the cooling water to flow once through the condenser loop is about 50 seconds.

From the tonnage capacity of the chiller and using manufacturer data, we calculated the heat transmitted from the refrigerant in the condenser to the cooling water in one pass.

From this, we figured how long it would take the cooling water to reach a safe inlet temperature of 55°F. It turned out to be less than 15 minutes. This gave us enough confidence to convince management the 3-way diversion valve was not necessary. This took some doing as management was apprehensive of going "against the grain" by bucking the manufacturer's recommendations.

However, they agreed we could go ahead with testing to validate our preliminary calculations. We also got the manufacturer's cooperation in testing our premise to prove our case one way or the other.

We tested our assumptions and calculations using cooling water at 39°F. There were no trips or malfunctions, and the vendor has concurred the 3-way valve is unnecessary.

However, the testing at the manufacturer's facility may not fully simulate installed conditions at our site. We have determined our cooling water loop is larger than that tested and, if anything, will help as the time of one pass through the condenser increases.

The chiller is now installing and will be operational later this summer.

The dividing line between process design and I&C Design is blurred and rightly so as it forces both groups
to work together to come up with the optimum solution.

However, in some cases (as in this case) examining the "process" requires expert knowledge of chillers and refrigeration, which are specialties in their own right and not necessarily available in a nuclear utility like ours.

It is difficult to get an off-the-shelf calculation or simulation plug-and-play package, especially for a complex system like this.

The alternative of a tailor-made package would require training, possibly additional vendors, and integration with the base package.

To avoid delays, cost overruns, and the like, a combination of hand calculations and validation testing may be the best approach.

It also possible to say this "incremental" approach, i.e. hand calculations, followed by testing, gets us closer to the physics and working of the system, which could be invaluable in finding faults.

This approach may not be suitable in all cases, and one's particular process has to see rigorous study before deciding on a simulation or control strategy.

This is another example of how reviewing system controllability adds value to process design.


Ken Dias (ken.dias@opg.com) is senior design engineer at Ontario Power Generation. The company has three nuclear, five fossil, and 64 hydroelectric generating stations in Canada.



Chiller is a machine that removes heat from a liquid via a vapor-compression or absorption refrigeration cycle. Most often water is chilled, but this water may also contain 20% glycol and corrosion inhibitors; other fluids like thin oils work as well.

Cooling towers are tower-like devices in which atmospheric air circulates and cools warm water, generally by direct contact-evaporation.

Cascade loop: Cascade control uses the output of the primary controller as the set point of the secondary controller as if it were the final control element.


PI control: The PID controller calculation (algorithm) involves three separate parameters-the Proportional, the Integral, and Derivative values. The Proportional value determines the reaction to the current error; the Integral determines the reaction based on the sum of recent errors; and the Derivative determines the reaction to the rate at which the error has been changing. A PID controller will be called a PI, PD, P, or I controller in the absence of the respective control actions. PI controllers are particularly common, since derivative action is very sensitive to measurement noise.

I&C: Instrumentation and Control