The effect of various control actions on anaerobic digester performance.

  • Kevin Monson

    Student thesis: Doctoral Thesis


    The sugar industry processes sugar cane and sugar beet to manufacture edible sugar. A high rate anaerobic system followed by an activated sludge process to 'polish' the effluent is presently the best available technology not entailing excessive costs for the treatment of wastewater from sugar processing factories.
    Upflow anaerobic sludge blanket (UASB) reactor systems are the most commonly implemented, with the higher loaded expanded granular sludge bed (EGSB) type systems gradually replacing at least some of the UASB applications. This higher loading carries with it a greater risk of process instability, especially given the irregular quantity and composition of sugar processing wastes. Control actions are required for process efficiency, preferably with appropriate control parameters monitored on-line. The need to balance economics (with regards to quantities of chemicals dosed) with the safety / stability of the process further exacerbates the need for close control. There are a limited number of control
    actions available e.g. bicarbonate alkalinity (BA) dosing, temporary diversion of load by feed rate variation, or dilution with final effluent.

    Monitoring of conventional anaerobic digester systems typically consists of a combination of intermittent manual sampling followed by off-line analysis coupled with qualitative observations. Hence a decline in reactor performance
    could go unnoticed for a significant period of time. A simple yet reliable control system could not only minimise labour, but could also react to any changes in reactor conditions as soon as they occurred, leading to more efficient process operation, higher quality treatment and ultimately greater economic gain. The
    industrially recognised 'bottleneck' in the quest for automatic control is the availability of reliable and cheap on-line analysers.

    In the work presented here a series of experiments have been carried out on a simulated sugar processing wastewater using an on-line BA monitor in conjunction with an adaptive control strategy developed in a parallel PhD project. The possibility of achieving successful automatic control of an EGSB reactor
    through its start up phase and also during steady state operation (including some degree of process optimisation) and a series of organic step-change experiments was investigated. Two different control actions to maintain a BA set-point in the reactor, organic loading rate (OLR) variation and BA dosing, were compared.
    Research was carried out on a 30 1 EGSB reactor rig operating at 37 °C, fitted with on-line sensors to measure temperature, gas production, carbon dioxide percentage and pH. All sensors were interfaced with a PC configured to scan the sensors at one minute intervals.

    Five attempts to automatically start-up the EGSB reactor were made (Experiments 1 - 5), using OLR variation as a control action at a constant hydraulic retention time (HRT) of 23.3 hours, with various improvements and adjustments made to the reactor and control system after each. It was concluded that it was not possible to automatically control start-up to steady state using OLR as a control action using the BA monitor and controller in their present forms. Main reasons for this were the susceptibility of the laboratory-scale BA monitor to blockage by biomass washed out of the reactor (a common occurrence during the start-up phase) and the severity of loading rate oscillations.

    For Experiment 6 BA was dosed according to the relationship of the on-line BA monitor output to the BA setpoint, and OLR was changed approximately weekly according to operator expertise based on on- and off-line data and visual observations. HRT was maintained at 22 hours. A successful, sustainable startup
    was achieved, with mean % COD removal during the first 10 weeks being 78 % at a mean OLR of 9 kgCOD/m3 /day. After the successful start-up period, the reactor's HRT was approximately halved to 11.2 hours, and two OLR stepchange experiments (from 10.0 kgCOD/m 3/day to 28.1 kgCOD/m3 /day, and from
    11.8 kgCOD/m3/day to 32.4 kgCOD/mVday) of twelve hour duration were carried out, followed by a removal of the control system and a similar organic step-change experiment (13.0 kgCOD/m 3/day to 32.9 kgCOD/m 3/day).

    It was found that approximately halving the HRT had no significant effect on the biomass or biomass activity. The HRT change did however adversely affect the smoothness of control, although control was not lost, as the on-line BA was always kept between 1500 and 2000 mgCaCO3/l. All parameters measured (pH, BA TVFA, effluent COD, carbon dioxide percentage, off-line methane percentage) indicated that conditions were less severe during and after organic step-changes when control was present. The controller also minimised the time spent at pH values potentially damaging to the bacteria (time spent at pH less than 6.0 in the two OLR step-change experiments where control was present (Experiments 6.3 and 6.4) was no greater than one hour, and in the experiment with no control (Experiment 6.5) was 8 hours) and returned the reactor to conditions conducive to efficient waste water treatment faster than when no control was present.

    Throughout all experiments controller oscillations remained severe. In this case, BA dosing as a control action was preferable, as the destructive effect of severe oscillations in the volume of BA dosed was considerably less than the effect of severe oscillations in the OLR, which repeatedly led to washout.

    Although it was possible to control the reactor subjected to OLR step-changes using the BA monitor based control system, control of the start-up phase was problematic. The BA monitor was not reliable or robust enough to provide the requisite data for use in the automatic control of the start-up of a high rate anaerobic digester. A thorough professional re-engineering of the B A monitor to deal with a greater sample flow rate, deliverable by a full-scale reactor (using wider bore tubing and larger flows, a more precise pumping system, and possibly with a suitable low maintenance sample filtration unit) could provide a suitably
    reliable and robust instrument.
    Date of AwardJun 2004
    Original languageEnglish
    SupervisorFreda Hawkes (Supervisor), Giuliano Premier (Supervisor) & Richard Dinsdale (Supervisor)


    • Anaerobic bacteria sewage
    • Purification
    • Anaerobic treatment

    Cite this