In this series we have been mentally exercising on control issues, and more specifically on the control of a chilled water application with cooling, on one hand, by a reciprocating compressor having four steps of capacity control. On the other hand, we looked at achieving this amount of cooling, and setting up control accordingly, with four scroll compressors being applied this time around. Early in this series we looked at control achieved from a temperature sensor placed in the water approaching the chiller, and then later at the circumstances where the temperature sensor was to be placed in the leaving water.
Our method was to base the thoughts on an electromechanical step controller, for the simple reason that we could picture the motor-driven cams being turned, and the micro-switches, with their cam-following fingers and little wheels riding up and down to initiate their switching actions. This way, the opening or closing of the switches on a pre-determined basis in response to the controller’s commands was how we achieved the programmed control. The cardinal issue so far has been that control of such a set-up may be based on either the measurement of entering water temperature, or upon leaving water temperature. Both have their virtues and both have their downsides, as we carefully considered.
The critical warning was that the chosen arrangement had to be specifically configured. If we hauled the sensor from, say, its position in the entering water and instead placed it in the leaving water because that made ‘more sense’ and attempted some minor programming adjustments of the controller to ostensibly accommodate this move, this would produce chaotic operation, frequently leading to failed compressors. Unfortunately, not everyone has been able to recognise chaotic operation when they encounter it. Typically, a large compressor would start, load up, proceed to unload and then stop, all in the space of a few minutes. Witnesses would remain unmoved. I am able to relate several personal experiences where exactly this has happened. The big alarm bell was that the people involved in most cases never became aware of the nature of the major error that had been in-built. They would replace or have repaired the failed compressor, and often carry on exactly as before, only to be greatly dismayed by yet another compressor failure after a surprisingly short time.
I earnestly hope all our readers have grasped the subtlety of this perilous situation, and will never get caught out in the way of so many of our unfortunate predecessors. While, to this point it has been alright to lump four individual scroll compressors into the same ‘block’ as those of a single reciprocating compressor with multiple capacity stages, the fact is that it’s not really acceptable. There is indeed a further difference.
The issue is, the reciprocating compressor itself wouldn’t be stopping and starting for every change in required capacity. The machine would have some form of capacity controller, sometimes hydraulically actuated and most commonly responsive to operation of a solenoid valve. This function would determine whether an operating piston and cylinder combination should be rendered either active or idle. Usually the targeted suction valves are physically held open while in the ‘unloaded’ situation, thus disallowing compression in the cylinder or cylinder pair for this period. Such an arrangement may load and unload repeatedly at very short intervals without experiencing any harm.
Not so, a standard scroll compressor – which stops or starts to respond to a requirement to match a change in the load of the moment with a change in capacity requirement – must provide a warning signal with each step change brought about by a motor either starting or stopping. It is with this thought in mind that we move on from where we left off last month. We will think of common failures in scroll compressors.
What lurks behind failures?
In my various exchanges on failures of scroll compressor difficulties, it has become increasingly apparent that a compact sequence of short runs offers a huge threat to the well-being of such a compressor. The rationale that underlies this is well worth further emphasis. Across those first seconds of operation which follow any start, such a compressor will throw an amount of oil. The anticipation is that, as the machine continues to operate over the next several minutes, this displaced oil will creep through the evaporator and then progressively migrate back to the compressor by way of the suction line. In this way, a normalised situation is quickly restored in the oil charge.
But, if there is no worthwhile ensuing run following the start, and the compressor is very soon again stopped, that displaced oil will be temporarily lost from the compressor. One or two such cycles possibly will not displace enough oil to cause the compressor any grief. The trouble will come if circumstances militate to result in this potentially harmful sequence consistently repeating across an extended period of time.
A useful solution could be to have one compressor of the group capacity controlled. Then the controller must have the capability of always repositioning operation of this compressor within the overall pecking order, so that it can provide the necessary in fills to offer infinite capacity control at whatever point it may be needed across the working capacity range of the entire group. In other words, while system load is in the range of zero to 25 percent, this compressor alone will run. But, if load rises to above 25 percent, a fixed capacity compressor will be started, and the variable capacity machine will back off to provide less than five percent of the installed capacity. This will be repeated as load climbs or reduces through all the programme steps. In this way, a relatively smooth ramp of capacity is offered throughout almost the entire range. It just requires much skill in applying the correct control hardware and adjustments. Those details, as best as I can configure them, will come next month. We first have other issues for discussion!
Figure 1 illustrates diagrammatically how we would hope to see this objective. To the best of my knowledge, there are just two ways by which this capacity variation function may be secured. The most obvious is to apply an inverter variable frequency control arrangement.
Application of an inverter makes it possible for a squirrel cage motor, which normally operates at a constant rotational speed, to be driven across a range of speeds. It does this by changing the frequency of the current being fed to the motor, on a controlled basis. But in most cases a compressor designed for fixed rotational speed is not suited to being subjected to this treatment. When a compressor has been manufactured specifically for such an application, it will probably have been designed to operate across the frequency range of 80 Hz down to 30 Hz, for example. It will be supplied with a suitable inverter as ‘part and parcel’ of the package.
But a difficulty can arise from the fact that the inverter is not likely to be as pristine as a Rolls Royce. Given ‘dirty current’, a ‘run-of-the-mill’ inverter is subject to a risk of failure. Back when I still had a day job, one of my periodic tasks was to record voltage of the electrical feed coming into some plant we might have had under review. In cases where a plant was suspect, I occasionally had to run across the roof of a building while dodging lightning bolts in order to set up the voltage recorder so we could later see how the installation was being affected by the storm.
Thunder storms in our parts, back when I was actively doing these things, seemed to occur considerably more frequently and with substantially more spectacle than is presently is the local case. When I was at school, we counted on getting two hefty thunder storms for every three days throughout the month of February. While I still live near enough to the same spot, an entire February can now come and go without presenting us with a single thunder storm. Under stormy conditions, it was common to obtain a recording very much as has been simulated in Figure 2. This would be garnished with voltage spikes, indicating how the electrical feed had been affected when there had been lightning strikes nearby. And those recorded spikes were plentiful across the hour or so duration of the storm. It would play havoc with some of our plants, although with the rugged and unsophisticated equipment of the times, seldom causing anything worse than a ‘nuisance trip’.
Copeland’s ‘digital scroll’
From what I am told, it is ‘ordinary’ inverters that can succumb to a feed which has become ‘spikey’, and can occasionally go belly-up for this reason. This susceptibility is the reason Copeland state it offers its ‘digital scroll’. They claim a machine using Copeland technology costs less than does an inverter-controlled machine, while offering greater capacity range and durability than the inverter option. It will be productive for us to take a look at Copeland’s technology in this regard.
Figure 3 illustrates the concept of the digital scroll. In the standard scroll compressor, the scroll tips of the orbiting and fixed scrolls are pressed into constant snug contact against the inner surfaces of the opposite scroll. Thus the machine provides fixed capacity when capacity is viewed in terms of the variables of suction and delivery pressure. Apart from going to a variable speed machine, there is a possibility that only a digital scroll to offer the means to alter capacity to match changes in refrigeration load.
The modified components applied to achieve this include first and foremost a fixed impeller which has been provided with the ability to be minutely floated up and down with a travel of a millimetre or so; just enough to free up contact between the scroll tips of the fixed and orbiting impellers. If we picture the entire situation here, there is concentric compression taking place as gas is ushered inward by compressor operation along its spiral path of diminishing volume from its two opposite points of entry at the periphery, to its ultimate compressed state at the central delivery port. For this to be achieved, snug contact is required between the scroll tips and their opposite contact points throughout orbital operation. However if this contact is broken, even when a clearance as small as just a millimetre occurs, pressure will equalise right across the impeller. As there is a check valve, or a one-way valve or non-return valve if you prefer, located in the delivery port, high pressure gas that has already been delivered is prevented from rushing back into the impeller.
How this is achieved
The fixed impeller has been equipped with an arrangement which contains some of the genes of a hydraulic relay. At the head of affairs is a form of piston. Compressed vapour is fed by way of a restrictive drilling, and is free to distribute across the face of the piston. Figure 3 makes a valiant attempt to show this.
There is also a bleed arrangement from the cylinder of this sub-arrangement. This bleed passes through a solenoid valve, and then returns to the suction line. Two things can happen:
If the solenoid valve is closed, there is no draining away of the feed of gas that has passed through the small drilling from the high pressure side of the scroll. This pressurised gas spreads to fill the piston area, causing the arrangement to be thrust down. This closes the contact of the scroll tips, and the compressor functions normally. This closure is done by way of a ring of compression springs, so as to get the contact pressure ‘just right’.
However, if the solenoid valve is opened, vapour in this leg will freely vent to the low pressure environment of the suction line. As vapour is being freely bled from the downstream side of the restrictor drilling, the entire region will fall to a pressure almost as low as suction pressure. With the previous down-force now having been removed, the impeller will be free to rise, pressed upward by gas pressure from below. The small clearance we have discussed is now realised at the scroll tips. Gas which is in transit through the impeller promptly bleeds across the scroll tips from high to low. The check valve closes, and low pressure prevails throughout the impeller. Notwithstanding continued operation of the impeller, there now will be no compression or delivery of vapour to the high side. The practically free-wheeling motor will consume minimal power.
Thus, by having a control system that determines whether the solenoid valve is to be activated or de-activated governs when the machine will pump vapour and when it does not.
Figure 4 will move us forward. The ‘smart’ controller divides real time into ‘slots’ for purpose of the control. The duration of these ‘time slots’ will be automatically optimised at between 15 seconds and 30 seconds. The solenoid valve may remain closed throughout. It may be held open for the entire period, or it may switch between ‘open’ and ‘closed’ at any point within each time slot. This duration of each ‘open’ and then the ensuing ‘closed’ phase is tailored by the controller as its response to the instantaneous load. The impeller may be commanded to deliver refrigerant for as little as 10 percent of the running time, and from there, anywhere up to 100 percent. The machine cannot be called upon to work at beneath 10 percent of its capacity, for its motor will be continuing to generate heat, and it is essential that this be transferred onward in a feed of refrigerant to travel to the condenser for due release from the system. The delay of these few seconds of ‘non-pumping time’ is however inconsequential.
‘Entering’ versus ‘leaving’ water control
In this series I have done my utmost to make plain the entirely different approach that is essential, according to whether ‘entering’ or ‘leaving’ water control is to be applied. It will probably have been clear from the discussions that it was the electromechanical controls of the past number of years that have been involved in occasional pandemonium, usually due to the sensing bulb of an entering control arrangement having been misguidedly placed in the leaving water, in the fictitious belief that this would ‘make the job work better’. The reverse would be equally bad, but it doesn’t seem as attractive to finger-happy folk.
But, to bridge into the next part of today’s story, we will consider an unfolding event that took place in a town less than a million miles from here. During an out-of-town training stint, I was asked to look in at a troublesome installation. It was a major installation, serving a water-based thermal storage arrangement. This installation incorporated four chillers, each equipped with its own pair of hefty reciprocating compressors.
The compressors serving these chillers were periodically suffering fatal trauma. As part of my free service for friends, I went to where dead compressors are mourned, and inspected the stator illustrated in Figure 5. The massive burn in its windings is apparent. What happens in many a basement plant is that the plant room is placed directly alongside the building’s transformer installation. The feed cable is therefore very short, making it possible for enormous amperage to flow in the event of a severe short circuit. Had this same machine been located in a rooftop plant room and undergone a corresponding measure of short circuit, the lengthy rising cable feeding the upstairs plant room would helpfully restrict the overfeed of current. The circuit breaker protecting the motor would trip, and upon subsequent repair, a somewhat smaller amount of damage would have been revealed.
Why should the rotor windings in this case have suffered such a traumatic fault? Say maximum running current rating had been 400 amps. The circuit breaker would have been set to trip at 400 amps after a short delay at which cut-out amps would have been greater to accommodate the current inrush associated with starting. Primary protection is the affair of the motor starter, and we are presently talking of the back-up current protection. In the event of a burn, this back-up protection would come to the rescue, so as to limit the extent of damage. This ‘back-up protection’ might comprise a suitably rated three-phase circuit breaker, or it could comprise fuses.
In this case, the installation had a circuit breaker for each circuit. But this circuit breaker would only be able to protect against a certain maximum current inrush. Say it was rated to protect against a current flow of up to 2 000 amps. If we had the upstairs plant room scenario with its lengthy cable feed, this could perhaps have limited the current rush of this particular case to 1 500 amps. The 400 amp breaker would have successfully tripped, protecting the associated motor switchgear and limiting the event to a minor burn.
When in a basement
Given a corresponding setup, but in the scenario of a basement plant room with its switchboard positioned close to the transformer of the main incomer, the same fault may well have yielded a current flow of 3 000 amps. The breaker would have tripped, but would have drawn arcs spanning the air gaps, allowing this 3 000 amp fault current to continue to flow. Dependence would shift to a protective device further up stream, and therefore with a considerably higher amperage rating. Now, instead of being rapidly quenched, the fault could persist even for minutes and longer, providing adequate opportunity for producing the advanced damage we see in Figure 5.
A High Rupturing Capacity (HRC) fuse has been inset in Figure 5. When applied in the circuitry, such fuses make it impossible for an arc to draw in the event of severe overload. The outer shell is of a hollow ceramic material. This contains the copper strip which forms the fuse element passing through the centre. This is completely immersed in suitable fine sand. Should the fuse rupture, the sand rushes in to occupy the void left by the vaporised copper. This instantly quenches the arc that would have formed, immediately halting the electrical feed. The damage that would have occurred in the Figure 5 stator would have been far less, although the motor will still have required a full rewind. In the absence of such protection, contactors would have welded closed, and you could have bet your boots the switchgear serving the Figure 5 motor would have been reduced to molten scrapheap copper by the event. HRC fuses would have probably protected the bulk of the electrical gear from such catastrophic failure.
A present dilemma
Although it’s informative, it is not in itself of specific importance with regard to our present dilemma. As mentioned, this was a chilled water thermal storage installation. And as we have said, feedback is absent in a case where the water being processed through the chiller might last have seen action some hours previously. The object of the exercise is to fill the thermal storage tank with water at 5ºC. Therefore the consulting engineer had rightly specified leaving water temperature control.
Not only that, with those earlier advances of technology some years back, all services were required to ‘keep up with the Joneses’ and have everything that could be controlled be controlled from a magic state-of-the-art computerised Building Management System (BMS), stashed into a PC. Therefore the original control arrangements that were part and parcel of the original chillers were orphaned by some deft snipping with a pair of side cutters. The cyber brain would now be invested in part with that control authority of the complete chiller installation. In compliance with the specification, that would be in terms of leaving water control. The standard electromechanical control, with which each machine has been imported, was indeed conformed for entering water control. But these controls were destined to serve out their lives as never-used museum pieces in the panels of the individual chillers.
I looked at the stator of Figure 5, and also at the near-by pile of scrap metal that constituted the earthly remains of the compressor that had most recently died. I had been standing in the wrong queue when ‘computer geek’ qualifications were being issued, and therefore know next to nothing of software and its performance. However, the detritus carried emblazoned all over it the all-too-familiar message which declared ‘Entering water control has been applied with a leaving water sensor’. This had to be it. Presumably the software writer had dug up an algorithm for ‘chiller control’ from some or other source. As a non-expert in air conditioning matters, he or she was unaware of the thrust of those critical issues we are presently discussing. The entering control routine had its ‘temperature’ function therefore orchestrated from a leaving water temperature transmitter. The settings parameters would have been subjected to ‘creative’ treatment to provide some hint of respectability.
This reportedly had by no means been the first such failure of compressors within this plant. But it would appear that, till now, the diagnosis had been “Well, having compressors fail is life for you”, and the butcher’s picnic was to continue unabated. While no follow-up report was ever favoured to me, I imagine the software specialist sought and found a P + I + D algorithm that incorporated a dead zone. Someone worked out the smallest workable value for the dead zone from compressor capacity control data, and the plant has run from that day to this with no further hiccup.
The reason for these considerations
What has been our reason for having travelled the latter part of today’s road? This has been to illustrate my belief that a competent computer programmer is not by any means also a polished refrigeration expert. And I believe that to have been a great problem of that era of a decade or so back. There are excellent new control types presently available that we could not even have dreamed about back during that period. We shall speak of a few of these. Their software is truly phenomenal, thoroughly tested and is not in any sense of the word ‘home made’. It is excellently documented, and brings the achievement of outstanding control success well within reach.
But that will take us across a fair amount of additional territory, and will aid us in extending the line of thought initiated by our Figure 1. Preparing this next article took me on a visit to my old friend from many years back: the one and only Pravin Kumar. He provided me with a considerable measure of guidance. But we already have had a full house for today. So we shall reserve those discussions concerning what Pravin passed on to me for next time. All being well, this will be in October. I look forward to us again being together then.