Why Sequencers Have Problems and How to Do Them Right
Tim Dugan, P.E. President, Compression Engineering Corporation
As readers of this publication know, there are many ways to save energy in industrial compressed air systems. One common supply side technology implemented is a “sequencer.” These can provide cost-effective savings. Unfortunately, many of them are turned off, or are not running properly. The goals of this article are to show why sequencers often have problems, and to demonstrate how avoid problems by proper system integration and controls design.
Introduction to Sequencers:
Sequencers are control systems that sequentially stage multiple industrial compressor systems, running only the minimum number required, based on one pressure signal, usually with only one running in a part-load mode (“trim”) and the rest either fully loaded (“base-load”) or off. In this article, we are describing three basic types of sequencers based on their algorithm, “cascade”, “target”, and “custom”. The first two are for “discrete” control only, using binary or relay interface, best suited for load-unload screw or reciprocating compressors. Custom sequencers can be applied to proportional control, which includes variable speed (VS) and centrifugal compressors.
The simplest sequencers use a “cascade algorithm”. It is the sequential starting and loading of compressors based on falling pressure, and the reverse for rising pressure. This algorithm comes from the pre-computer age. Sequencers started their life as a mechanically-driven pressure switch selectors, using relays, cams and timers. They work like this: as pressure drops, the next compressor starts and loads, and then the next starts and loads if pressure drops further. As pressure rises, the reverse occurs. The last on will load and unload once the number of compressors running stabilizes. The sequencer swaps the order around to even out wear. This was coded into simple programmed logic when programmable logic controllers (PLCs) and embedded controllers were introduced to industry. The cascade algorithm is best suited for positive displacement and reciprocating compressors. Cascade sequencers have a wide operating pressure differential.
With the advent of PLC and embedded controller technology, different algorithms have been designed. One common alternative to the cascade algorithm is the “target” algorithm. There are variants, but the simplest uses one pressure band for the trim compressor and a wider pressure band to trigger base-load compressors. The sequencer manages the number of base-load compressors running without having to wait for pressure to continue to drop again. The first time the pressure drops to the lower “base-load” point, the trim compressor is already fully loaded and the #1 base starts. The second time it hits the same base-load point, the #2 starts, and so on. The reverse happens at the high limit of the wider pressure band, but in reverse. Another way is to use timers instead of the wider pressure band to determine if the next compressor needs to start. Target sequencers have a narrow operating pressure differential.
These common designs seem simple. In a “perfect” world, implementation would be simple also. However, simple sequencers described above assume the following system characteristics for smooth implementation:
All of the compressors are the same vintage, make, type, and size.
All are in the same location
All can run load-unload and be remotely started and stopped
All are plumbed to a common header, generously sized, preferably before dyers
There is adequate storage.
Unfortunately, the systems that they are being implemented in often don’t look like that.
The sky is the limit here, but we will comment briefly on three algorithms:
"Flow-based”. The optimal number and size of base-load compressors are run at any time based on total flow, not strict sequential order. A more advanced processor and algorithm is required. Significantly different size or type of base-load compressors make this algorithm a good fit.
“Load-sharing”. Multiple proportional compressors are run at the same pressure and percent load. The management system “bumps” the local settings to make this happen. This expands the effective range of efficient trim operation, making a system more stable and reducing blow-off. Multiple (3 or more) centrifugal compressors of similar size are a good fit for this algorithm.
“Hybrid base-trim”. Trim compressor(s) are run by either a cascade or target algorithm, sometimes at an elevated pressure and behind a pressure-flow controller. Base-load compressors are controlled by a separate algorithm, and run at a lower pressure, often final system pressure. A good fit would be a mix of centrifugals and screws.
Common Sequencer Problems:
1. Improper algorithm for the situation. For instance, cascade control for centrifugal compressors. This all but guarantees that one centrifugal will be in blow-off most of the time, wasting about 75% of its full load power.
2. Lack of a champion, leading to controls being defeated.
3. Control logic and wiring not changed when equipment is changed, particularly compressors.
4. Not all compressors are operating in “auto” properly. This can be caused by:
Incomplete or improper interface wiring and/or programming, most often on the compressor side.
Compressor is manually put in “local” for real or perceived reliability reasons.
5. Multiple compressors are in part load. Three common examples:
Incomplete integration: The minimum command is given to the compressor by the sequencer, “make air”. If the compressor is making little or no air because of local unloading or modulation, the sequencer will start and load the next compressor at part load. The sequencer is “unaware” of this.
Improper local setting: The local modulation settings are in the same range as the sequencer settings. The sequencer starts the compressor once, but pressure never gets high enough to unload it. The local settings reduce capacity first. Then another compressor ends up being called to start when demand increases. The sequencer is “unaware” of this also.
Compressors far apart: Local controls trigger a compressor to operate out of phase with master control input, which is far away from it.
6. Short cycling (rapid loading/unloading). This can cause oil carryover in oil flooded screw compressors and reliability problems with oil free screw and centrifugal. It has a variety of causes:
Operating pressure differential too tight.
Inadequate control storage, causing rapid pump-up and bleed-down times.
Excessive pressure differential across treatment equipment. Effective compressor control band is reduced by the dynamic pressure drop across the treatment. This often is caused by a compressor being isolated behind an individual restrictive dryer and filter.
Different local and sequencer control points. The sequencer might be using pressure downstream from the dryer and the compressor unloads based on a pressure upstream.
7. Excessive motor starts. This can damage the motor. Besides the “short cycling” issues, it can be caused by:
Timers in sequencer either not adjusted properly or can’t be adjusted.
Sequencer not tuned properly.
8. Improper integration of a VS compressor. This can cause the following:
The VS compressor is controlled in a discrete manner, it could either base-load (full speed), shut off, or run uncontrolled, depending on the settings. All VS compressors control their own speed by their internal controls, so a simple sequencer is usually unaware of these problems.
The VS could hunt, “chasing” the compressor that is being loaded and unloaded by the sequencer. Lack of storage, tuning, or location of compressor could cause this.
Problems Unique to “Vendor Sequencers”:
Since these are designed and sold by compressor companies, they are usually intended for integration with all new compressors from the same manufacturer. They are typically based on proprietary code running on low cost embedded controllers, and usually use some form of either cascading or targeting. Some of the lower cost sequencers are what I call “told you to make air” controllers, with no feedback that the compressor is actually doing that. Most vendor sequencers don’t incorporate intelligent feedback for running, load or fault. Some incorporate monitoring, but only points that are already monitored by the compressor controller. They are often low cost (less than $10,000) and appear simpler to implement than other options. Some unique problems are:
Sequencer is sold as a component on a project, with no integration service. Underselling is perfect way to have under-performance.
The compressors don’t have the same vintage control panels and interfaces. Oftentimes, these sequencers require an up to date controller on all the compressors, or the interface becomes quite primitive. You either get all the interface or just a single “told you to make air” contact. Not much in between.
The algorithm is not known to the field engineer and customer. Proprietary controllers try to be too smart. Unfortunately, if you don’t know its basic logic you can’t get it adjusted properly.
Problems Unique to “Third-party Sequencers”:
These are sequencers that are designed and sold by third parties, often PLC-based open architecture. They have a pre-programmed algorithm that is adapted to each site. They typically have relay interface or a generic network interface that has to be programmed or adapted in each case. Some are customizable for customer networks, data collection, etc. They are typically mid cost ($10,000 to $25,000 before customization). They usually come with a field engineer start-up included. Some unique problems are:
Supplier might not understand compressor issues well. Their strength in controls knowledge can be offset by their lower understanding of the compressors themselves.
Sequencer might not be able to interface with compressor controller’s full communications capability.
Supplier has little local support after project is completed. It is made by a small company that has a thinly-stretched field engineering staff.
How to “Do Sequencers Right”
Here are our recommendations for a successful “sequencer” project that continues to work for the life of the system:
1. Don’t call it a “sequencer”. That implies it is a simple add-on component, which it certainly is not. It is an integration project that happens to have master controls as a part of it. A better term for it is a “compressed air management system”.
2. Identify a project “champion” in-house who has some understanding of controls and of the compressed air system.
3. Select a controls architecture that is best suited to manage the compressed air system. Keep in mind that compressed air is an essential utility that is very expensive to operate, and even more expensive if it fails. Skimping on controls and particularly monitoring can end up being a bad decision in the long run.
4. Perform a system audit first. Assess the existing system in as much detail as is cost-effective. We recommend the following:
Data-log compressor room primary variables for about 1-2 weeks.
Calculate total compressed air flow profile, total power, and system efficiency (acfm/kW).
Develop several alternative project concepts, from retrofit controls to new equipment, depending on budget.
Develop a preliminary compressed air management system specification.
Some compressor vendors are qualified to do this. Independent auditors have significant value as well.
5. Select the best firm to design and install the compressed air management system based on the architecture issues and audit results.
6. Develop approval drawings for the management system, and review them. This should include a written sequence of logic.
7. Select a contractor to install the management system. They need to be capable of working closely with the auditor, local electrician/engineer, management system supplier, and compressor vendor. If other mechanical issues are being done at the same time (storage, piping, dryers), consider making them a subcontractor to one turn-key contractor.
8. Get compressor interfaces identified and modified first. Test them to make sure they are all ready to be controlled by the management system.
9. After approval, build and program the management system.
10. Have all mechanical and instrumentation issues completed and tested.
11. Deliver the management system.
12. Land wires and/or network, test interfaces, and start up the system. It should be run through failure modes and exception modes sufficient to tune the system.
13. Collect data for at least one week and deliver it to the management system supplier or auditor for review. If possible, allow them to have direct remote access to pull data, either through the plant HMI system or a GSM modem.
14. Develop a tuning/commissioning report based on this data.
15. Perform final tuning.
16. Document the system well. A three ring binder in the maintenance and system champion’s office. At a minimum, include an overall P&ID, electrical schematic(s) for controller and interface wiring, and written sequence of operation.
17. Hang an engraved sign on the wall near the controller with the approved set points.
18. Train the champion and operators.
The most important issues that affect compressed air management system performance are people and business related. Are the right people doing the right thing with the right technology at the right price in the right way and in the right sequence? Implement an integration project properly, and the sequencer will have a much better chance of working properly.