Accumulation returns n-times more value than actual number of summed values present. The factor n is dependent upon the number of times the parent object
updates are called before rule that does the summation is invoked in the flow.

How to reproduce?
A simple test project is attached to this report. This project has following DataModel.

InvokeServiceRequest=>LoanRequest==>TransactionDetail==>PolicySet and WorkingObjectTransactionDetail

<?xml version="1.0" encoding="UTF-8"?>
<process xmlns="http://drools.org/drools-5.0/process"
xmlns:xs="http://www.w3.org/2001/XMLSchema-instance"
xs:schemaLocation="http://drools.org/drools-5.0/process drools-processes-5.0.xsd"
type="RuleFlow" name="RF6365_200" id="278865" package-name="com.test.sumofall.GeneratedRules" version="200" >

<header>
<imports>
<import name="com.test.sumofall.InvokeServicesRequest" />
<import name="com.test.sumofall.TransactionDetail" />
<import name="com.test.sumofall.LoanRequest" />
<import name="com.test.sumofall.WorkingObjectsTransactionDetail" />
</imports>
</header>

<nodes>
<ruleSet id="1" name="1" x="307" y="31" width="80" height="48" ruleFlowGroup="group_1" />
<start id="70004" name="Start" x="61" y="33" width="48" height="48" />
<ruleSet id="2" name="2" x="463" y="27" width="80" height="48" ruleFlowGroup="group_2" />
<ruleSet id="3" name="3" x="315" y="230" width="80" height="48" ruleFlowGroup="group_3" />
<join id="90004" name="EndJoin90004" x="532" y="80" width="49" height="49" type="2" />
<split id="11" name="11" x="337" y="101" width="49" height="49" type="2" >
<constraints>
<constraint toNodeId="90004" toType="DROOLS_DEFAULT" name="ServiceFlowCode Equal to Post Rules Processing" priority="1" type="rule" dialect="mvel" >$invokeServices :
InvokeServicesRequest( )

$lendingTransaction :
LoanRequest( parentId == $invokeServices.myId )

$transactionDetail :
TransactionDetail( parentId == $lendingTransaction.myId )

$workingTransactionDetail :
WorkingObjectsTransactionDetail( parentId == $transactionDetail.myId
, serviceFlowCode == "Post Rules Processing" )</constraint>
<constraint toNodeId="3" toType="DROOLS_DEFAULT" name="ServiceFlowCode Not Equal to Post Rules Processing" priority="1" type="rule" dialect="mvel" >$invokeServices :
InvokeServicesRequest( )

$lendingTransaction :
LoanRequest( parentId == $invokeServices.myId )

$transactionDetail :
TransactionDetail( parentId == $lendingTransaction.myId )

$workingTransactionDetail :
WorkingObjectsTransactionDetail( parentId == $transactionDetail.myId
, serviceFlowCode != "Post Rules Processing" )</constraint>
</constraints>
</split>
<end id="80004" name="End" x="640" y="30" width="48" height="48" />
</nodes>

<connections>
<connection from="70004" to="1" />
<connection from="1" to="2" />
<connection from="2" to="11" />
<!--If WE change order of the following two lines to this it works for this example but similar change doesn't work when multiple split and joins are present in ruleflow
<connection from="11" to="3" />
<connection from="11" to="90004" />
-->
<connection from="11" to="90004" />
<connection from="11" to="3" />

<connection from="3" to="90004" />
<connection from="90004" to="80004" />
</connections>

</process>

***ROOT CAUSE***
I believe the problem is caused by the childLeftTuple.getStagedType() being equal to LeftTuple.INSERT instead of LeftTuple.NONE or LeftTuple.UPDATE.

Thread [main] (Suspended)
PhreakJoinNode.updateChildLeftTuple(LeftTuple, LeftTupleSets, LeftTupleSets) line: 443
PhreakJoinNode.doRightUpdatesProcessChildren(LeftTuple, LeftTuple, RightTuple, LeftTupleSets, ContextEntry[], BetaConstraints, LeftTupleSink, FastIterator, LeftTupleSets) line: 363
PhreakJoinNode.doRightUpdates(JoinNode, LeftTupleSink, BetaMemory, InternalWorkingMemory, RightTupleSets, LeftTupleSets, LeftTupleSets) line: 307
PhreakJoinNode.doNode(JoinNode, LeftTupleSink, BetaMemory, InternalWorkingMemory, LeftTupleSets, LeftTupleSets, LeftTupleSets) line: 48
RuleNetworkEvaluator.switchOnDoBetaNode(NetworkNode, LeftTupleSets, InternalWorkingMemory, LeftTupleSets, LeftTupleSets, LeftTupleSinkNode, BetaMemory, AccumulateNode$AccumulateMemory) line: 547
RuleNetworkEvaluator.evalBetaNode(LeftInputAdapterNode, PathMemory, NetworkNode, Memory, SegmentMemory[], int, LeftTupleSets, InternalWorkingMemory, LinkedList<StackEntry>, LinkedList<StackEntry>, Set<String>, boolean, RuleExecutor, LeftTupleSets, LeftTupleSets, LeftTupleSinkNode) line: 533
RuleNetworkEvaluator.innerEval(LeftInputAdapterNode, PathMemory, NetworkNode, long, Memory, SegmentMemory[], int, LeftTupleSets, InternalWorkingMemory, LinkedList<StackEntry>, LinkedList<StackEntry>, Set<String>, boolean, RuleExecutor) line: 334
RuleNetworkEvaluator.outerEval(LeftInputAdapterNode, PathMemory, NetworkNode, long, Memory, SegmentMemory[], int, LeftTupleSets, InternalWorkingMemory, LinkedList<StackEntry>, LinkedList<StackEntry>, Set<String>, boolean, RuleExecutor) line: 161
RuleNetworkEvaluator.evaluateNetwork(PathMemory, LinkedList<StackEntry>, RuleExecutor, InternalWorkingMemory) line: 116
RuleExecutor.evaluateNetwork(InternalWorkingMemory) line: 54
DefaultAgenda.evaluateEagerList() line: 958
DefaultAgenda.fireNextItem(AgendaFilter, int, int) line: 913
DefaultAgenda.fireAllRules(AgendaFilter, int) line: 1198
StatefulKnowledgeSessionImpl.fireAllRules(AgendaFilter, int) line: 1202
StatefulKnowledgeSessionImpl.fireAllRules() line: 1175
SumOfAllTest.main(String[]) line: 133