Plant Operation Series: A practical guide to backwash tuning for reliable sand filter operation

“Follow these proven practical backwash tuning techniques to improve your sand filters' reliability and robustness”

Raw Water Treatment Basics and Pain Points

Water is a fundamental part of refining, petrochemicals, and power industries. Some examples of its applications include:

1.    Makeup water for an open recirculating cooling water system

2.    Boiler feedwater for steam production

3.    Firewater

4.    Etc.

As a first step, raw water must be treated or clarified before being used as makeup water in most utility processes. Clarification and filtration remove suspended and partially dissolved solids, bacteria, and other forms of impurities to help prevent system scale, corrosion, and fouling.

A simplified flow diagram of a typical raw water treatment plant is illustrated here as an example in Figure 1.

Figure 1 A simplified raw water treatment process

Clarification takes place in three sequential steps as followingly described:

1.    Coagulation. Coagulants, e.g., inorganic salts of aluminum or iron, are usually added to accomplish coagulation. As coagulation prefers high shear and rapid mix, coagulants are generally injected into the feed line upstream of a static mixer to induce rapid mixing.

2.    Flocculation. Flocculants, i.e., high-molecular-weight, water-soluble organic polymers, are typically added to enhance small particles' agglomeration into larger particles. Flocculation ideally takes place in an environment with low shear, high retention time, and moderate mixing. Typically, flocculants are directly injected into a mixing zone of a clarifier.

3.    Sedimentation. The physical removal of agglomerated particles from suspension or settling, usually under low water velocity in a clarifier's settling zone.

 In addition to coagulants and flocculants, disinfectants (such as Sodium Hypochlorite) are also applied to suppress microbial growth. On a rare occasion, pH adjusters, i.e., Sodium Hydroxide, can also be used for neutralization.

There are multiple types of clarifiers available in the industry. A sludge blanket clarifier is only presented here as one example, as per Figure 2.

Figure 2 A sludge blanket clarifier (Courtesy of the Permutit Company, Inc.)

In most raw water clarification, at least a portion of the clarified water is filtered by the aid of a sludge blanket in the settling zone. Clarifier effluents of 2-10 NTU may be improved to 0.1-1.0 NTU by conventional sand filtration.

Generally, raw water clarification is a highly-manual process and often requires intensive human monitoring. Consequently, it is prone to process upsets. For example, a sludge blanket might disappear from excessive sludge draining and subsequently causes high turbidity in effluent water.

When this happens, a downstream sand filtration system is the last resort to ensure acceptable suspended solids concentrations in the finished water, thus requiring extra care. In this article, the author will outline proven tuning and troubleshooting techniques to improve your sand filters' reliability and robustness.

It should be noted that pressure sand filters will be used as examples here, but one can also apply the main ideas to other types of sand filters.

Sand Filter Operation

Pressure sand filters are typically filled with multiple filter media, e.g., sand or anthracite, to enable deep bed filtration mode. The use of different filter media is the main reason why they are also called multimedia filters. As presented in Figure 3, anthracite with a larger diameter is placed on top of smaller diameter sand to trap larger-sized flocs/particulates, following by gravels as support. As a typical design standard, flow distributors are equipped to aid flow distribution in all operating steps.

Figure 3 A standard sand filter’s internals and simplified loading scheme (Courtesy of Serck Separation Technologies)

Table 1 summarizes the standard size and density of filter media used in multilayer filtration. For most designs, a filtration system consists of multiple filter vessels to allow periodic backwash without filtration operation being interrupted.

Table 1 Media used in multilayer filtration (adapted from Ref. 2)

As time progresses, the pressure drop across these filter vessels will gradually increase. Once the pressure drop reaches the maximum limit (typically 1.0-1.2 bar for pressure filters), these filter vessels need to be cleaned (backwashed) to remove trapped particulates, so they can again return into service. Some operators use a turbidity breakthrough as the main criterion, while others just set a fixed-timer to trigger backwash operation, e.g., every 24 hours.

Most designs are relatively simple, as illustrated in Figure 4. They usually employ a set of automatic on/off valves to enable each operating step given as an example in Table 2, along with other peripheral equipment, e.g., backwash pump. Generally, a PLC controls these operating sequences, e.g., which valves to be open/close or when the backwash pump shall be started/stopped. 

Figure 4 A typical sand filter setup. Only one vessel is outlined here as all vessels are identical.

Table 2 Typical Operating Steps

Routinely, these filter media need to be fluidized and expanded to a certain extent by a counter-current backwash flow for cleaning. In general, the degree of bed expansion during any backwash operation should be greater than 10% for efficient cleaning but less than 20% to prevent filter media loss. According to industrial surveys, poor backwashing is the most frequent cause of filter failures.

In addition to the degree of bed expansion, the design backwash rate also depends on the size and density of filter media but generally falls in the range of 15-25 m/h. It is very important to use filter media with proper size and density to ensure the validity of the design backwash rate.

Evaluate your backwash performance

During your field test, if backwash water is relatively clean at the start with insignificant turbidity increase throughout the backwash operation, it often means the backwash flow rate is inadequate to clean filter media. On the contrary, if there is a significant amount of anthracite (or sand) in backwash water, the backwash flow rate is too high.

Figure 5 shows real backwash profiles of two different backwash operations. The blue-colored trend implies that the backwash operation was highly efficient and completed after 10 minutes. Additionally, the turbidity changes in a backwash cycle happened abruptly, with a significant peak in the second minute.

On the other hand, the red-colored trend implies an inefficient backwash operation as the turbidity profile was relatively flat. Regardless of the backwash duration, this filter vessel would never be thoroughly cleaned.

As a general criterion, American Water Works Association (AWWA) has recommended terminating backwash when the turbidity is in the range of 10–15 NTU (Ref. 1). In addition to water loss, excessively long backwashing is also detrimental to the filtration process, contributing to post-backwash turbidity breakthroughs and requiring an extended ripening period.

Figure 5 Efficient vs Inefficient Backwash, with 1 minute sample frequency

Positioning Screw, Valve Opening and Backwash Flow Rate

Many of these systems do not have a dedicated flowmeter for backwash flow monitoring, especially old or budgetary units. Most of them rely on the valve opening of these on/off valves. As illustrated in Figure 6, both positioning screws (circled) can be varied up to a certain degree to regulate the backwash flow rate. In other words, these on/off valves serve two purposes in this type of application:

1.    As an on/off valve for sequential control 

2.    As a fixed-opening control valve for flow regulation.

Figure 6 Valve opening can be adjusted by varying positioning screws highlighted by red circles (Courtesy of Tameson)

The valve actuator's nominal rotational angle in Figure 6 is 90 ° but can be varied between 80° to 100° by altering the insertion length of positioning screws. Some actuators have a more comprehensive range of rotational angles than the model used as an example here.

In addition to on/off valves, some units install dedicated regulating valves to adjust the backwash flow rate. As a preventive measure, these regulating valves should be locked with a car seal once the backwash flow rate is tuned to prevent accidental adjustment.  

Problems associated with valve malfunction

The functionality of these on/off valves is crucial to a successful operation. Their rotational movement should be smooth, while the seat tightness should be acceptable.

A frequent maintenance program is a key to success but at the same time could also be a burden.

As part of preventive maintenance, these valves are periodically disassembled, repaired, and tested at a local maintenance workshop. Sometimes, an operator might observe malfunctions and request valve maintenance, which usually means more or less the same maintenance steps to be performed.

Regarding the author’s experience, these good intentions often turned into a disaster. Many instrument technicians do not realize how important it is to have a positioning screw optimally adjusted. They just follow the standard maintenance procedure and put these valves back in place.

Quite often, the insertion length of these positioning screws deviates from their optimal position after valve maintenance. More often, the engineer in charge does not even notice if the valve was removed for repair, as an operator or shift supervisor is typically the one who gives maintenance permission.

For a well-equipped system, there will probably be a flow meter for backwash flow monitoring. After the valve returns into service, a board operator or an engineer can monitor from a control room whether the backwash flow rate deviates from the original value. If you are fortunate enough, the backwash flow rate would be the same as before, thus no need for further adjustment. Conversely, the field adjustment would be required if there are any significant deviations.

As earlier mentioned, these water treatment units are mostly budgetary. Some units are just outdated and poorly designed, especially when they are a support unit, i.e., as part of a refinery. Back then, the author used to operate both a luxuriously designed hydrocracker and a very poorly designed water treatment system; both units belong to the same refinery. More often than people think, there is no flow meter available for backwash flow monitoring.

Failure to detect these anomalies can lead to performance issues of filtration systems. In one instance, some filter vessels hit the maximum allowable pressure drop in a short period after each backwash. To identify the problem, the author then went to the site for investigation and found a significant amount of anthracite in samples taken during backwash operations. These observations implied that these filter vessels had already lost their deep bed filtration capability from excessive anthracite loss. The filtration mechanism had become cake filtration, where most of the carried-over sludge accumulated at the top of the sand layer rather than between voids of an anthracite bed, causing premature plugging and a sharp rise in pressure drop. In this particular case, there is no flow meter installed to monitor backwash operation.

Similarly, an inadequate backwash flow also results in premature plugging of filter media, and it could be as bad as shown in Figure 7. The pressure drop across filter vessels will sharply increase over a short time and result in an unacceptably short cycle length with possible premature turbidity breakthrough. Additionally, too frequent backwash operation resulting from premature plugging often causes water balance issues. On several occasions, the author could even tell that the backwash flow rate was way too low by just observing backwash water flowing into the underground drain.

Figure 7 A severely plugged multimedia filter. A top surface of anthracite was fully covered by carried-over sludge from an upstream clarifier

Apart from the issues mentioned above, an excessively high-pressure drop can also lead to internal damages. For example, the air distributor's welding seam shown in Figure 8 was cracked due to the chronic high-pressure drop, thus creating a gap allowing anthracite and sand to enter then plug air passages. The occurrence of air distributor damage rendered air scouring ineffective.

Figure 8 A damaged and severely plugged air distributor with a gap highlighted by the red circle

After observing a significant amount of anthracite in backwash water, the author then inspected every on/off valve associated with the backwash operation (valve no. 7 and 8 in Figure 4). As per the author’s inspection, some positioning screws were misplaced, with too short insertion length. In other words, these valves had an overly wide opening, which in turn resulted in an excessively high backwash flow rate and losses of anthracite.

A decision has been accordingly made to reload new filter media into these filter vessels. As expected, the author observed a nearly complete loss of anthracite upon inspection. This finding was in line with the fact that these filter vessels had already lost their deep bed filtration capability.

Upon completion of reloading, these newly loaded filter vessels were then commissioned. As previously pointed out, the insertion length of these positioning screws was misplaced and needed to be re-adjusted. The backwash flow adjustment had become a very challenging task when there was no flow meter available.

One might think about using non-intrusive flow sensors, e.g., ultrasonic flow meter. That is a good option providing there is an adequate pipe straight-run. Unfortunately, these filter vessels and backwash systems are usually located close to each other and often in a cramped space. As illustrated in Figure 9, the piping systems typically consist of many bends and relatively short pipe sections, rendering the ultrasonic flow meter useless. Some piping materials, such as low-density plastics and rubber (or even thick paint), can attenuate signals or possibly create false echoes, causing inaccurate measurements.

Figure 9 A typical layout of sand filters with bends and short pipe sections (Courtesy of Serck Separation Technologies)

A trial and error could also be carried out for a single filter vessel to determine the optimal backwash flow rate, e.g., by evaluating the turbidity of backwash water. Although, it is not a piece of cake regarding the author’s experience. Even if the optimal backwash flow rate is eventually determined, a more critical question is how to ensure the same backwash flow rate for every filter vessel. The piping systems connecting these filter vessels are mostly asymmetric. Consequently, the backwash flow rate of each filter vessel will not be the same despite all backwash inlet valves (valve no. 7 in Figure 4) have the same opening.

A practical guide to backwash tuning

One technique that the author successfully applied in the past is estimating the backwash flow rate by the pump characteristic curve. This technique has been proven to be effective and consistent on many occasions.

If there is no flow meter available on site, one may consider following recommendations given below

1.    Have a clear flow rate target (Q) in mind, then determine a corresponding differential head (?H) from the pump characteristic curve, as illustrated in Figure 10.

Figure 10 A graphical demonstration of ?Hdesign determination

The author recommends using the design backwash flow rate (Qdesign) as a starting point.

This corresponding ?H can be converted into a differential pressure target across the main pump using the following formula.

Where,

Pdischarge, a pressure reading of the pressure gauge located at the discharge of the main pump

Psuction, a pressure reading of the pressure gauge located at the discharge of the spare pump

ρwater, water density

g, gravitational acceleration

For more understanding of how to use field data, consider the system depicted in Figure 11. The main pump is in service delivering water for a backwash operation. In this particular case, Pdischarge is a pressure reading of PG1, while Psuction is a pressure reading of PG2. In other words, PG2 is an indirect indication of the suction pressure of the main pump, as indicated by the green isobaric zone highlighted in Figure 11.

Figure 11 A simplified illustration of a typical backwash system

If there is no spare pump available, just read a pressure gauge located at the discharge line to estimate Psuction before commencing the backwash operation. As per the author’s experience, this is an acceptable assumption and usually good enough for practical purposes.

2.    Try to tune the backwash flow rate by adjusting the insertion length of corresponding positioning screws (valve no. 7 or 8 in Figure 4). A good strategy is to start at a low flow rate to prevent filter media loss, e.g., at 50% of the full insertion length, then gradually increase until the estimated backwash flow rate matches the design value. Another advantage of this tuning strategy is that corresponding positioning screws will not be pushed against the instrument air pressure, making the field adjustment much more effortless. For most cases the author has come across, a design value is good enough to achieve efficient backwashing.

3.    After finishing a preliminary adjustment, the author highly recommends repeating the backwash performance test earlier described. If the performance is still unsatisfactory, consider increasing the backwash flow rate by 10%. The author does not recommend going beyond 10% as it might result in excessive loss of filter media. In case the backwash performance is still unsatisfactory after increasing the backwash flow rate, the real root cause of inefficient backwash operation should be identified, e.g., an air scouring step might be ineffective.

4.    Seasonal temperature variations can be substantial in some areas. The colder the water is, the more filter media will expand at the same backwash flow rate. The author recommends tuning the backwash flow rate at the design value during the warm season and periodically checking for filter media loss during the cold season, i.e., by observing backwash water. These seasonal variations are another reason why the author does not recommend adjusting the backwash flow rate beyond 10% of its design value (a typical design factor). Table 3 provides correction factors for standard sand and anthracite at different temperatures as a starting point for tuning.

Table 3 Correction factors for typical filter media at various temperatures (adapted from Ref. 1)

5.    To ensure good backwashing over a comprehensive range of water levels in the storage tank, tune the backwash flow rate when the water level is at the minimum, then observe filter media losses at the maximum controlled water level (graphically illustrated in Figure 12). Ideally, there should be minimal anthracite (or sand) loss over the entire controlled range. In the case of excessive media loss at the maximum water level, the author recommends raising the minimum level and readjusting the backwash flow rate, so it will not increase too much at the maximum water level. 

Figure 12 A recommended tuning strategy for varying water level

6.    If it fails to reach the design flow rate even at the full valve opening, check for other possible issues. Figure 13 shows an example where fine dirt/particles almost 100% blocked the underdrain distributor.

Figure 13 A blocked underdrain distributor. This underdrain distributor has too narrow water passages making it susceptible to plugging and extremely difficult to unblock.

Example

On a hot summer day in a refinery in South East Asia, an engineer tried to optimize the backwash flow rate for four identical dual-media filter vessels containing anthracite and sand. From a design reference book, he found that the design backwash flow rate (Q design) is 300 m3/h at 30 °C. His water treatment plant was suffering from high-pressure drop across multimedia filters. Some operators and shift supervisors had already started calling him a moron recently, mainly from his failures to solve the problem.

Unfortunately, this refinery is too poor to install a flow meter for backwash flow monitoring, leaving this engineer no choice but to use the pump characteristic curve in Figure 10 and field pressure data to estimate the backwash flow rate. The backwash system this engineer was dealing with is similar to what presented in Figure 11. It should be noted here that he learned this practical tuning technique from an excellent Linkedin blog about everything chemical engineering he came across the other day.

1.    Practical tuning

From Figure 10, the corresponding differential head (?H) for 300 m3/hr of backwash flow is 20 m. As it was a hot day, the water temperature was approximately 30 °C with ρwater of 996 kg/m3. 

Which means:

This calculated differential pressure will be a target for the field tuning.

To ensure the pump curve's validity, he first confirmed whether both backwash pumps are still in good condition with the refinery's maintenance department. Everything from actual blade size to failure record was thoroughly reviewed and accepted. Therefore, the pump characteristic curve illustrated in Figure 10 should provide a good flow rate estimation.

Before commencing the backwash operation, the positioning screw of the backwash inlet valve (valve no. 7 in Figure 4) was pre-adjusted at approximately 50% of the full insertion length as a preventive measure to prevent excessive filter media loss.

Regarding initial field pressure data, he found that Pdischarge-Psuction was 2.50 bar (approximately 150 m3/h of backwash flow) while the target is 1.95 bar. Consequently, he requested an instrument technician to adjust the screw length, so the differential pressure became as targeted, approximately 1.95 bar.

More than that, a visual check of backwash water also confirmed that the backwash flow rate was too low to expand the filter media as the water appearance was quite clear. Once the differential pressure approached 1.95 bar, a significant amount of sludge and silt had been observed in the backwash water, with a negligible amount of anthracite. A dirty appearance of backwash water implied an efficient backwash operation. It was also proof that the differential pressure target of 1.95 bar somehow worked, although a full backwash performance test would still be required.

2.    Ensure robustness against water level fluctuations

On the day this engineer adjusted the backwash flow rate, he requested the board operator to reduce the water level in the storage tank to 85%, the minimum level.

To ensure the system's robustness, he also monitored the backwash operation when the water level was at 95%, the maximum level.

As per the field monitoring, he found minimal anthracite in backwash water with the estimated backwash flow rate of 310 m3/h (or 3% above design) using the same technique described earlier. Ideally, the maximum/minimum water level should be properly controlled to ensure efficient backwash operation and minimal media loss.

3.    Minimize risks of media loss from temperature variations

At 85% water level and 30°C, the backwash flow rate is 300 m3/h. This system utilizes 1.0 mm anthracite and 0.5 mm sand.

Statistically, the temperature could drop significantly to 20°C (ρwater is 998 kg/m3), thus requiring a lower backwash flow rate. From Table 3, the adjusting factor for 1.0 mm anthracite is 0.91/1.10 or 0.83. In the same fashion, the adjusting factor for 0.5 mm sand is 0.90/1.13 or 0.80.

The author recommends using the lower adjusting factor for correction, as followingly exampled:

Backwash Flow Rate @ 20°C = 0.80 x 300 m3/h = 240 m3/h

From the pump characteristic curve in Figure 10, the corresponding ?H is 15.5 m water.

Thus,

When the water temperature is 20°C, one can first use 1.52 bar pump differential pressure as a first guess; then observe the real backwash and adjust as needed.

Closing thoughts

·      For poorly-designed or budgetary existing systems, where there is no flow meter available for backwash flow monitoring, the outlined tuning technique has been proven to be highly effective and provide consistent backwash flow rate on many occasions.

·      The author recommends installing a dedicated flowmeter for the new unit or retrofitting the existing plant for early detection of any anomalies. Although it still requires a field check to ensure problem-free backwash operation, the proposed technique is nonetheless useful to prove the accuracy of your flow meter.

·      Create a transparent logging system to ensure everyone involved in process monitoring can track the status of valve maintenance, e.g., the unit engineer can create a dedicated section in the shift report for any critical valves currently under maintenance.

·      Communicate with all instrument technicians in charge to ensure they fully understand the effects of positioning screws on filter operation.

·      For troublesome clarifiers, where sludge carryover is frequent, the author recommends setting the backwash flow rate at 10% above design value to handle potential clarifier’s upsets. One may consider installing a real-time sludge level detector for better sludge level monitoring.

·      Create a task for a field operator in charge to monitor the backwash operation, at least once a month for every filter.

·      Seasonally or quarterly conduct backwash performance evaluation.

·      Be careful when extending a backwash interval. A backwash flow rate or duration should be accordingly adjusted to ensure efficient backwashing.

·      The same tuning technique can also be applied to other systems with similar configurations, e.g., activated carbon filter or sand filters in waste-water treatment.

Reference

1.    Filter Maintenance and Operations Guidance Manual, AWWA Research Foundation

2.    Handbook of Industrial Water Treatment, Suez