Influent composition effects on sludge particle size distributions in an MBR
T. Maere*, G. Bellandi*, M. Lousada-Ferreira**, L. De Temmerman*, J.W. Mulder**, I. Nopens*
*BIOMATH, Department of Mathematical Modelling, Statistics and Bioinformatics, Ghent University,
Coupure Links 563, B-9000 Gent, Belgium (, , , , Tel: +3292645937, Fax: +3292646220)
**Evides Industriewater, Schaardijk 150, 3063 NH Rotterdam, The Netherlands (, )
Abstract: The effect of influent composition and operational conditions on MBR performance and sludge filterability was investigated at labscale,aiming at improving theperformance of the full-scale MBR of Terneuzen (the Netherlands). Hereto, alab-scale MBR treating real wastewater was run at similar conditions to the full-scale MBRand subjected to well-defined operational changes: a) addition of effluent coming from aconventional activated sludge (CAS) wastewater treatment plant, b) addition of leachate from a landfill, c) addition of effluent coming from a waste reclamation plant. The performance of the MBR was checked on multiple levels, i.e. biological performance, membrane fouling and sludge quality, and the measurements encompassed normal influent, effluent and sludge parameters, next to transmembrane pressure, standardized filtration tests and particle size distributions. The results showed minor effects of operational changes on TMP valuesand the standardized filtration test values only increased when having the combined effect of adding 30% CAS effluent, 2.5 % leachate and 1.5 % waste reclamation effluent. On the other hand, all operational changes caused consistent and measurable shifts in particle size distributions, measured by two different methods: laser obscuration and image analysis.Though the results look promising, further research is required to evaluate the suitability of each of the aforementioned applied methods.
Keywords: Membrane bioreactor; filtration performance; operational effects
Introduction
The full-scale MBR “De Drie Ambachten” in Terneuzen (The Netherlands) is operated by Evides Industriewater NVto providedemineralisedwater to industry after further treatment steps. At times, when the MBR influent flow is not sufficient to satisfy the water demand,additional effluent from the nearby conventional activated sludge treatment plant is fed to the MBR. In a quest for optimized process performance and better operational conditionsthe question arose if this modus operandicould be a cause ofsludge filterability issues and/ormembrane permeability decline. Both hypotheses were investigated in this study by running a lab-scale MBR at Ghent University under similar conditions as the full-scale installation.
Materials and methods
BioreactorTSS concentration / 6-8 g l-1
DO setpoint (intermittent aeration) / 2 mg l-1
Temperature setpoint / 15 °C
X-flow airlift membrane module
TSS concentration / 7-10 g l-1
Cross flow velocity (50-50 air/water) / 1 m s-1
Average filtration flux / 31.8 LMH
Backwash flux / 106 LMH
Influent (carbon dosage)
Rule 1: minimum COD content / 100 mg l-1
Rule 2: COD substrate addition limit / 50%
Rule 3: BOD5/TN setpoint / 3
A detailed description of the used lab set-up can be found in Jiang et al. (2009). A reference state was established after running the lab-scaleMBR for two months on wastewater coming from the conventional wastewater treatment plant (WWTP) of Destelbergen (Aquafin NV, Belgium)at fixed operational settings that mimic the normal operational conditions of the full-scale MBR plant in Terneuzen (see Table 1).After this period the operation was changed wherebythe effluent from the WWTP of Destelbergenwas used as 30% of the MBR input and later even 60%.Special influent streams, i.e. leachate from a landfill site and effluent from a waste reclamation plant (denoted as wrpe from now on)which account respectively for up to 2.5 % and 1.5 % of the total influent flow of the full-scale MBR in Terneuzen, were tested as well. The experimental schedule is given in Table 2.
Measurements entailed the characterisation of normal influent, effluent and sludge parameters (COD, BOD5, TN, NH4, NO3, TP and TSS) next to standardised sludge filtration tests (Thiemig, 2012) and particle size distributions (PSD - EyeTech, Ankersmid).
Table 2: Experimental schedule of the lab-scale MBR
Phase 1 / 1 - Start up / 9 Jan - 20 March2 - Reference period / 20 March - 27 March
Phase 2 / 1 - 30 % CAS effluent + 70 % influent / 27 March - 13 May
2 - 50 % CAS effluent + 50 % influent / 13 May - 15 May
3 - 60 % CAS effluent + 40 % influent / 15 May - 29 May
4 - 30 % CAS effluent + 70 % influent / 29 May - 3 June
Phase 3 / 1 - 28.75 % CAS effl. + 68.75 % infl. + 2.5 % leachate / 3 June - 21 June
2 - 29.25 % CAS effl. + 69.25 % infl. + 1.5 % wrpe / 21 June - 1 July
3 - 28 % CAS effl. + 68 % infl. + 2.5 % leach. + 1.5 % wrpe / 1 July - 10 July
Results and discussion
MBR sludge and influent measurements:The TSS measurements in Figure 1 show reasonably stable sludge concentrations, although the influence of varying influent concentrations during the study period is noticeable. The influent measurements (Figures 2, 3, 4) show variability in the composition of the influent. In general, the influent can be considered weak and this is even more evident during rainy weather. The latter, combined with influent degradation during the week (fresh influent was taken on a weekly basis), makes that continuous addition of extra carbon (see Table 1) was absolutely necessary.
Figure 1: TSS measurements MBR
Phase 2 of the experimental campaign consisted of adding various amounts of secondary effluent from the conventional activated sludge (CAS) treatment plant of Destelbergen to the raw sewage influent of the lab-scale MBR, similar to the actual operation of the Terneuzen MBR. The CAS effluent had a quite constant composition but some degradation was occurring during the week as shown in Figure 4.
In Phase 3 of the experimental campaign leachate as well as waste reclamation effluent was added in quantities comparable to the real situation at the MBR in Terneuzen. The composition of these special substrates is shown in Table 3.
Figure 2: Weekly fresh influent composition (total COD, soluble COD, BOD5)
Figure 3: Weekly fresh influent composition (TN, NH4, TSS)
Figure 4: Average influent (left) and CAS effluent (right) degradation over the course of a week
Table 3: Special influents composition
Total COD / Soluble COD / Total N / Total P / NH4 / NO3 / BOD5 / pHLeachate / 2660 / 2720 / 1181 / 19.6 / 0.176 / 20.7 / 424 / 8.61
Wrpe / 1138 / 1042 / 82.1 / 5.01 / 0.337 / 7.49 / 45.1 / 7.56
Nutrient removal and membrane filtration results: The results in Figure 5 show reasonable effluent nitrogen levels up to the start of phase 3 of the experimental campaign. The effect of leachate addition is clearly visible from the NO3 levels but also in the COD results. The effect of waste reclamation plant effluent addition does not show as much effect in the MBR effluent, meaning that it might be degraded and/or retained by the membrane. The BOD5 values in Table 3 point however to the latter option. The wrpe substrate also caused foaming of the sludge. In general the effluent COD values are rather high compared to influent COD levels. However, a large fraction of the influent COD is inert (see Figure 4).
Figure 5: MBR effluent composition (COD, TN, NH4, NO3)
The transmembrane pressure (TMP) and filtration test values in Figure 6 and 7 point towards an easy filterable sludge at the start. A first clear increase in TMP values is visible in the period of 60 % effluent addition. However, the real onset of the increase coincides exactly with a change of the aquarium pumps used for bioreactor tank mixing. The pumps might have caused an increased breakage of sludge flocs. However, a second time, around the end of June, where the aquarium pumps were cleaned (not changed) did not result in a sudden increase of TMP values. The increased TMP diminished when the 60 % effluent addition scenario stopped, possibly indicating that the 60 % effluent addition had an effect as well. The standardised filtration test values showed a small increase in the middle of the 60 % effluent addition period and then dropped again.
An increased rate of fouling, but certainly not a sudden increase, was visible after running some time with leachate, leading to the only chemical membrane cleaning required during 6 month of operation. However, when looking at these results it remains a question if the increase was a result of operating the lab-scale MBR with leachate addition, or if it was just the accumulation of several months of fouling. Especially since the filtration test values didn’t show any increase.
The waste reclamation effluent addition didnot cause major changes in TMP or filtration test values, although the sludge was foaming heavily. On the other hand, a major increase in filtration test values was observed at the end of the project when leachate and waste reclamation effluent were dosed together. Interestingly this effect was not visible in the TMP values, indicating that the used standardized filtration test is not exactly mimicking real sludge filtration conditions.
Figure 6: Standardised filtration test according to Thiemig (2012)
Figure 7: Complete TMP profile
Particle size distribution results:Given the considerable amount of PSD curves that were collected, only a few of the particle size distribution (PSD) measurements, i.e. laser obscuration and image analysis resultson samples of bioreactor sludge, are given in Figures 8 and 9. The results are presented as percentage of deviation from the reference distribution collected in the reference week (represented by the horizontal axis at 0 %). A difference was observed between the two measurement techniques (laser obscuration or image analysis) as well the way of processing the data (volume or number based, not shown).
The results in Figure 8 indicate the biggest difference in PSD for the 60 % CAS effluent addition scenario whereby particles around 1 µm seem to increase significantly. As mentioned before, it was also in this period that the pumps for bioreactor mixing were changed. The results for leachate addition do not seem very different than for the normal 30 % CAS effluent addition case. The results for wrpe addition (and foaming sludge) do not deviate much from the reference distribution.
Figure 8: Particle Size Analysis by laser obscuration - bioreactor sludge - number based distributions
Somewhat different than above, the results in Figure 9 show the biggest changes for the leachate and wrpe addition and not for the CAS effluent addition scenarios. The effect of leachate is still more pronounced than for wrpe. The differences in PSD consistently occur in the lower and middle range of particle sizes, which is different than for the results of Figure 8 where changes only occurred in the lower range. The change in X axis scale compared to previous results should be noted. The results show an increase in small particles and decrease of middle sized particles (i.e. sludge flocs).
Figure 9: Particle Size Analysis by image analysis - bioreactor sludge - number based distributions
The PSD results of the CAS effluent (not shown) indicated that the particles present in this CAS effluent might be the cause (to some extent) of the small particle increase as observed in Figure 8. This finding is important as the increase in the amount of small particles may form an indication of worse filterable sludge. However, at the modest filtration conditions under which the lab-scale MBR was running it was difficult to verify this statement. It cannot be excluded that for more challenging settings a worse membrane performance would be noticeable.
The cause of sludge floc degradation as observed in Figure 9is likely related to the special composition of the leachate and wrpe and could be a cause of sludge filterability issues, as was apparent at the end of the experimental campaign.
Conclusions and perspectives
The effect of influent composition and operational changes on MBR performance was investigatedona lab-scale MBR and resulted in the following:
- A sudden increases in TMP only occurred once when the bioreactor mixing pumps were changed. The effect of different operational scenarios on membrane fouling was not clearly visible from the TMP results. However, it cannot be excluded that for more challenging settings a worse membrane performance would be noticeable.
-The standardised filtration test values showed a big increase at the very end of the experiment when leachate and waste reclamation plant effluent were dosed simultaneously. However, as stated before, this was not visible in the TMP values.
-The particle size analysis showed shifts in PSD when operational conditions were changed. Large flocs (image analysis) seemed influenced by the addition of special influents: leachate and waste reclamation plant effluent. Small particles (laser obscuration) appeared to be influenced by CAS effluent addition (and maybe physical shearing of the mixing pumps).
If the experiment was to repeated the following details should be taking into account:
-The effect of mixing pump shear stress on floc breakage.
-Effect of more challenging filtration settings in the membrane loop, e.g. flux, crossflow, aeration, sludge concentrations.
-Physical meaning of the standardised filtration test.
-Effect of temperature.
References
Jiang, T., Sin, G., Spanjers, H., Nopens, I., Kennedy, M.D., van Der Meer, W., Futselaar, H., Amy, G., Vanrolleghem, P.A., 2009. Comparison of the Modeling Approach between Membrane Bioreactor and Conventional Activated Sludge Processes. Water Environment Research 81 (4), 432-440.
Thiemig, C., 2012. The importance of measuring the sludge filterability at an MBR - introduction of a new method. Water Science and Technology 66 (1), 9-14.
1